Google Gemini FULL COURSE 5 HOURS (Build & Automate Anything)

Google Gemini full course 5 hours build and automate anything. Today I'm going to show you the complete Google Gemini course that will change everything about how you use AI. This is a 5-hour compilation of everything I've tested, built, and discovered using Gemini over the past few months. Google just dropped massive updates that most people don't even know about. We're talking about building full apps, creating animated story books, setting up AI agents, and automating your entire business. All completely free. This is the most comprehensive Gemini training you'll find anywhere. By the end of this, you'll know more about Gemini than 99% of people using AI right now. We have some brand new updates from Google Gemini. And today I'm going to run you through exactly how they work, how to use them, etc. And these are completely free. So, one of the newest updates, for example, is the personalization. Also got a bunch of other methods. I've actually left a bunch of notes inside the AI profit boardroom if you want to learn exactly how to do it. So, one of the biggest updates here is the guided learning mode, right? So, this is basically like an interactive teaching mode and it basically tests you and shows you exactly how to learn things quickly. as similar to for example something like notebook LM which is a really powerful free tool as well but basically this is a really cool way to do guided learning as you're learning something throughout the journey right so let me show you an example of exactly how this works you can now use you've got an option between Gemini and guided learning in Gemini you see the difference here so there's two different ways of using it right we're going to be testing this for example we've got an example right here, right? And you can see here we've got I'm a freshman preparing for an exam in my intro in biology class. Can you help me learn about enzymes? Blah blah. And then we can use the guided learning method to help us learn faster. Now, at the same time, what we can also do is we can say, okay, recall my recent conversations with you. And you can actually ask Gemini now for questions and that sort of thing. And it will actually recall recent information, right? This is for example looks like we've had a few conversations recently. We discussed funny scenes with AI articles on the future of AI agents clustering keywords news headlines blah blah. Okay. And so it's really easy and powerful to do this and it doesn't take a long time but it is a bit of a game changer. Now if you're using something like chat GBT I think normally you do have to pay for that feature whereas now for example it's fully loaded with all of the different ways you can use this and it's super simple and easy to do as well step by step. All right. Now, Roz says, "Gemini 3, is Gemini 3 out? " It's not out yet, but hopefully it comes soon. So, I haven't seen Gemini 3 come out yet, but there are a lot of rumors and that sort of thing. Check out my other video on it if you want to learn exactly how to use it step by step. So, you can see here if we go back to Google Gemini now. So, to get a good handle on this free exam, we could focus on a few key areas. Can you quiz me? And then it should be able to actually turn this into an interactive quiz which is great because now you can basically just learn way faster, right? And so you can actually see what it's doing here is it's taken the problem that I had which was learn about enzymes. Absolutely don't miss out from my school days learn biology. And then from here you can see that we've got an enzymes quiz and we're living the enzyme dream my friends. All right. And then you can for example get tested. I've actually got a question right there. That is unbelievable isn't it? And then you can go through the quiz step by step like this directly here. What else we got here? So we got guided learning mode. We've also got temporary chats, right? So not only the personalization that we were talking about previously where you can recall recent things. Another cool thing you can do here is you can say okay knowing what you know about me, what's a big area of growth? This is the way that I used to chat actually trying to ask it like based on our conversations, based on what you know about me, etc. Where are some areas I can improve? what are some systems I could set up, etc. And then you can easily get this bad boy set up, right? And you can see here, for example, yeah, you have a strong creative interest in artificial intelligence. You're pretty weak at your enzymes for biology, mate. So, here's some other things you could actually combine the two based on our previous conversations. Does exploring this sound interesting to you? Blah blah. And then you can go off and start creating quizzes or whatever you want. Really simple and easy to do. So, let's keep going now. So, we've talked about two different things there. Number one is the personal context. Number two is the guided learning mode. Let's have a look at the temporary chats up here. So if we go to new chat, we've got a bunch of options. So we got import code. We should be able to start a new chat here. But before I get into that, I want to show you this as well. This is quite interesting. So we've got these story books, right? So basically, you can create a customized picture book for either children or adults. Give a topic and an optional target audience and an optional art style for images. Right? So let's plug this in. We'll test it out. So, I'm going to say, okay, based on the recent updates in the AI world, for example, chat GBT, give me a storyboard for adults that would be funny talking about the world of AI and how bad Chat

GBT 5 was despite us waiting 2 years for it. Something like that, right? And then we can actually start building out a story book using this process, right? So, you can see it's now creating a story book, writing the first draft. So you can actually like create audio books and story books on the spot for you. Pretty crazy. So we'll just wait for that to load and we should be good to go on that in a second. What else we got here? So whilst that's loading, we can come back to that in a minute. We also have image generation. Right now this is actually available apparently inside Google Docs as well. But also you can use the new image gen 4 that just came out inside here. Right. So if you go to image gen generate with image gen and then we'll say a cat eating biscety and we'll just wait for that to load and that will just generate an image. Right? So it's basically like an interactive sort of learning environment now where you can build apps, you can build quizzes, you can learn what you want and I actually think like Gemini Pro is easily one of the best out there. It's right up there in terms of responses versus Claude. Claude can't generate images and it definitely cannot generate pictures of cats eating biscatinos like this. that's world class. And then also you've got, for example, stuff like chat GBT5, but I know a lot of people feel let down by that and it's not quite as good. So Gemini might be a good option. I also think one of the strongest points of Gemini is that it can actually create really good apps directly inside here. Right? So if you ever been like, I want to build an app, but I don't know how to do it, we have a lot of training inside the AI profit boardroom, but also you can just go inside the chat here, toggle on canvas, right? So you can start a new chat and then from the canvas feature you can start building stuff out, right? So for example, let's go inside here and we'll say create an app about and then you plug in whatever your app is about. So let's say for example it is about maybe keyword research for SEO, right? And what it's going to do is just start going off and building that out. So wait for that in a minute. But in whilst that's coding out in the background over here, what we're going to do is we're going to go onto Gemini. And what you can see is you can actually if we unmute this, you can actually listen back to the storyboard, right? So you can not only flick through the storyboard and have a great time reading it, but also you can just listen to it. So let's have a listen to this. I feel bad for Barry. Barry's had a hard day in the office. This man look all excited about chat GPT5. Brenda's just not having it. She's not having it. Now if you don't have That's interesting, actually. So you might not be able to hear that back. So what I'm actually going to do is I'm going to change the speaker. One second here. — Let's see if you can hear this or not. Let me know in the comments if you can hear this or not. — The year is 2025. For two long years, the world held its breath for one thing. Chat GPT5. Tech gurus prophesied its arrival. Influencers vlogged about its potential. And Barry, a man whose greatest technical achievement was programming his coffee. — Can you hear that or not? If you can, just let me know in the comments so that I can help you. If it's not working, I will share a link inside the air profit boardroom. You could just steal that from me. But essentially, it's pretty amazing that you can just create a story on the fly and then you can go from there. So, I'm going to just put that inside the our profit boardroom here. — Waited most impatiently of all. This is Brenda, he'd say to his long-suffering wife, his eyes gleaming with the fervor of a true believer. It's going to write my novel, plan our vacation, and finally figure out what the day it launched. The internet practically buckled with trembling hands. Typed his first momentous query. Chat GPT5. What is the meaning of life? He leaned back, ready for enlightenment. The AI word pixels danced. And then the answer appeared. 42. He tried again. Okay. Can you help me write a best-selling sci-fi novel? Chat GPT5 responded instantly. Synopsis. A brave toaster and a cynical microwave go on a quest to find the legendary golden spatula. It's a buddy comedy with lasers. What about that weird smell in the fridge? He typed, his last hope dwindling. The AI's response was a single unhelpful image. A stock photo of a man shrugging. Brenda walked in holding a Tupperware container at arms length. I found the smell, she announced, her voice flat. It was your two week old science experiment with leftover chili. Barry looked from the sentient appliance novel plot to the shrugging man on his screen, then to the biohazard. his wife was holding. The great leap forward for humanity felt more like a clumsy stumble into a puddle of forgotten leftovers. He started to laugh. It was a slow rumbling chuckle that grew into a full-blown belly laugh. "Two years of hype, of breathless anticipation for an AI that was essentially a glorified taco obsessed magic eightball," he said, wiping a tear from his eye. "At least we know what's for dinner. " Brenda just shook her head, a small smile playing on her lips. The future was here and it was gloriously, hilariously, disappointingly human and it definitely wanted tacos. — All right, there peeps. I had to move the microphone so that's why I didn't talk. But essentially what you can see here is we have a generated video with

one single click. I literally generated it live for you today. The video was made fully with Gemini. The images were made with Gemini. It was all storyboarded and put together like an artifact style with audio and it's sharable. You can also flick between a lower pitched voice and a higher pitch voice," — he said, wiping a tear from his eye. "At least we know what's for dinner. " Brenda just shook her head, a small smile playing on her lips. — It reminds me a little bit of Blinkist, but slightly more amusing and interesting. All right, so let's click on share. Now, if we click on share, we can create that audio book. I can share it with the world. And then you've got that ready to go. Kind of like I've added down here, right? And that's a sharable link. And so you can do that with Gemini. Pretty cool. That was the audio generated right there. Some people asking like what are the export functions. So I think you can print this off, right? So you can actually download it as a PDF. I don't think video from what I can see. It's not really a video. It's kind of like more a PDF document with a voice over at the top. So I can imagine that's why they just make it sharable. Maybe as well they don't want people going out and spamming YouTube or something like that which totally makes sense. We don't want that. We don't want just a whole YouTube to be AI storyboarding. Steven, head of growth at Goldie Agency is here. He says, "This is me spying on my boss. What a legend. " Good to see you here, Stephen. Just perusing LinkedIn. If you saw me there, this is what I do on this is what we do for fun. If you're wondering what I'm doing, by the way, just so everyone's clear, Stephen works with our agencies. these absolutely awesome guide to work with and if you want to book in a sales call link in the comments description to book in a free SEO strategy session. All right, so we're good to go. We can download it. We can start from the beginning. We can listen in full. And literally all I said here, it was very creative. I just said based on the recent updates in AI world, for example, chat 5. Give me a story board for adults that would be funny talking about the world of AI and how bad Chachi 5 was despite us waiting 2 years for it. And if you're good with Stephen in the comments, he might even send you a video of him dancing at a Goldie Agency Poland get together that was absolutely legendary recently. All right, so we're going to kick this back off now. We've basically got Gemini creating storyboards. It's really good at creating personalized conversations. It creates these guided learning modes and it's very good at generating images. What other AI in the world can do this? Tell me. Literally just tell me one AI in the world that can do this apart from Gemini because the image quality is topnotch. You might get good outputs from Claude but you won't get good image quality. You could use chat GPT5 but can you create animated storyboards on chat GPT5? No. And everyone feels quite let down by it I think just in general. And then also you can create these quick SAS tools I call them. Like for example we can say here create an app about keyword research for SEO and then boom shack lacker. We can get that right here. So if we click on get suggestions, we've got all these keyword suggestions. We just built out a keyword tool that's actually sharable with people. And then we've got a bunch of keyword ideas here, right? And it was all just created in one shot. Looks really nice. And so you can see the power of Gemini. Like honestly, if I had to focus on one single if I said who's going to win out of all the AI options out there, who's going to win? I would genuinely say Google are probably leading the race. If you look at what for example chat are doing, they rely heavily on marketing, right? They got someone amazing like a frontman Sam W Mortman and he's a great frontman for the brand but at the same time if they didn't have that marketing then it would be a struggle right if you go back to Google now as well let's just check yeah Google absolutely smashing it right if you look at their growth over the last year or so they're absolutely flying they seem to be making very good moves in the industry in general right don't bear in mind as well they got stuff like nobook LM they've got stuff like for example Google jewels they got Google CLI etc So, it's really powerful stuff. Aaron says, "Google is too censored. Church is less restrict. " It depends what you're doing with it, right? For me, I don't really have anything that would be going against their guidelines. So, I guess I just don't come across that, but it depends what industry you're in, right? Depends what dodgy stuff you're doing on there. And Abdul says, "Please tell me how to use it shortly. Then I will see the live afterwards. " All right. So if you want to do this, all you do is you go to Gemini and then you know you say okay create a storyboard about and then you just type in whatever the storyboard is about. Right? So let's say for example SEO is dead because chat GPT came along and destroyed it. But we're happy now because actually people search more than ever. Actually, I'm going to use link whisper flow for this more than ever because AI search is growing every day as people get addicted to searching in there and because they can actually get better responses in Google. All right. By the way, if you're wondering what I just use for that's whisper flow. So, we're going to say create a storyboard for that. That should toggle it on. And then let's see if that triggers automatically or if I have to click one of the options. I think we're going to be good to go on that. It's generating an image, but I can't see a

story board. Let's go back to story book. A story book. All right. So, we have to do it in here. We just say okay create a story book and then that should toggle it on. All right so just to recap for the steps you go up to gems story book you say create a story book about whatever you want and then it will start creating that right and goes off and does that bad boy. What else we got here? So Veil 3 says I think Veo3 is supported in Gemini Studio. Can we convert the story book in video? As far as I'm aware you can't. What you can actually do is you just get something like dscript is really good or you can use something like for example streamlabs. You just record your screen. Even Loom would probably do the job. And then from there all you would do is you would just go back to the story book that we create. So like this one. And then once that's done just record your screen whilst you're playing it, right? And play it full screen. And then that's problem solved, right? You got the Google file ready to go. User X says, can you please talk about batch processing and scalability? how we can scale this type of applications. Yeah, you can use something like NA10, right? Just keeping it really simple, but if you want to keep this stuff simple, I assume if you want to scale applications, you can use something like NA10. It's really easy to just scale stuff and have it working on a repetitive loop with the API. The other option that you have, right, let's say, for example, with batch scaling, you're talking about clawed. So, you can use clawed code for that. I'm not going to get too much into it because it goes way outside the scope of this live stream, this video, but yeah, feel free to jump in the AR profit boardroom if you want more help on that. Right, if you go to the calendar here, we actually do five coaching calls a week. So, you can just jump on the coaching calls, ask us any questions live. If you can't make it, you can ask questions inside the community and that sort of thing as well. Let's see what we got. Google's prototype team is really good. The quality of AI products they are releasing is really useful. I'd 100% agree with that. Right. Bear in mind like we're only looking at the front end of Gemini. But if you go to AI Studio, it's redonkulous what else you can do with it, right? So for example, if you go inside AI Studio here, you can stream your screen in real time and just speak to the AI like what other AI can do that right now, right on the desktop. You can also, for example, generate media. You can, for example, build stuff. So you can build apps directly with Gemini here. And there's all these different apps that you can actually just test if you just want to have a go at something that's already pre-made and you're not sure what you're doing. And then also, yeah, you've got the build and the stream section. Let's go on the stream section here. So, if we go to share screen, and I'm just going to go to share screen here. We share screen, too. What's cracker lacking, my friend? — Not much. Just hanging out. — Okay, great. So, I'm going to go into Google Gemini. Could you give me an idea for an SEO app that we could build? — Sure. How about an app that analyzes website content for SEO optimization? And — all right, so you get the point. It was a little bit laggy there. I think it's cuz I'm live streaming, but you get the point. Like you can just ask it to come up with SEO optimization tips and then whatever. It's good to go. All right, living the dream right there. And then user X says, I was working on a master data management project for an industry. The problem is scalability and even if we do batch processing, we still lose a lot of context. Any suggestions, please? It goes way outside the scope of this. But I would just look more into batch processing and how you can do it. There's another way you can actually get the costs of for example claude down as well. But also bear in mind all these apps are all these APIs are upgrading in terms of their context windows. So for example claude this week increased its context window by 1 million, right? So it went from 200k to 1 million. Then you also got for example Google Gemini that's 1 million context window. And to be honest like they probably give it the next update. Who knows what Google 3. 0 is going to be like. Google Gemini 3. 0. It's going to be absolutely crazy. Sarang says there's a print error when it doesn't print. Okay. And then DT says, "Yo, this screen share feature is really solve a painoint for a lot of people. " Yeah. And crazy thing about this is that nobody even knows about it or even really understands how to use it properly, but it's an absolute game changer. So now we're going to listen to The Owl and the Firefly by Julian Goldie SEO apparently. Now if we click on print, let's have a look here. Will it print or not? So you can see it's loading the bar right there. Just coming back to question. It's preparing it. Is it going to break or is it not? It did something. It didn't download it. Let's click on print again. Yeah, you might be right there. Doesn't there might be a bug, but bear in mind it's like it's relatively new. Fair enough. So let's play this now. — The owl and the firefly. In the heart of the big bustling city of Wabopoulos lived an old owl named Silus. From his perch high at top the great library, he had watched the city grow. For years, everyone came to Silas for answers. He knew all the secret paths and hidden alleys of Obopouolis, the best keywords to whisper to get the best results. He was the master of search engine optimization, or SEO for short. Then one day, a new light appeared in Wopoulos. It was Sparky, a tiny firefly who zipped and zoomed through the city streets, leaving a trail of brilliant light. Sparky didn't know the old paths. He made new ones. He didn't whisper keywords. He had conversations. He was an AI and he was changing everything.

Soon the crowds the — Just coming back to this. So Abdul says, "What is this about? How can we use the storyboard? " So essentially what you can do with the storyboard is you can create like interactive storyboards to learn about stuff. A lot of people for example quite a few people inside the air are profitable born and have asked like how can you create comics or books etc. And this is one of the best ways to do it. One of the easiest And essentially what this book is about. This is more honestly this is more like a I would say it feels more like a kids story to be honest. If I go back let's X off that. Yeah. So we said create a story book about SEO instead blah blah. I think it's created more like a kids style story but you could ask it for an adult story as well. So for example then you can come up with a bunch of ideas like all right. So if we go to recent as well. So you can see recent and then we can listen back to this one. User X says, "Can you talk about image recognition and video recognition capabilities of Gemini multimodal capabilities? " Everyone talks about image and video generation real world industry. Yeah. So, so what you can do for example when you're looking at for example like video generation, right? Video generation is really cool because you can create if we go back here if I click on let's see what we got video right and you can actually add a v photo. So you can actually animate yourself as well. Let's pull up a photo of me and then we can actually use that to generate a video. So bear with me folks. I've downloaded a photo and then we can upload that for consistency into Gemini. All right. So we're going to upload a photo now. Let's take that picture. And then from here we're going to say, right, okay, create a video of this man telling a funny joke and then flying into the sky. Something like that. So we're taking that picture. And the reason that it's pretty good for uploading the picture is you can basically create character consistency. So previously if you were generating a video every time you asked it to generate say image of a man it would be like a different man every time or even if you were very specific with the description it would change every time. With this process instead you're going to have the same character inside each video that you generate. Right? And so now we can generate the video. This is using BO3 which is by far one of the best ways to do this. And then also if we went back to let's start another chat on the side right we went back to another chat here and then we took an image right so we'll upload an image we'll say okay based on that multimodal capability that image recognition for example we'll say describe this person and then it should start analyzing that picture and analyzing me as a person so just and this is the other cool thing as well is you can do this in different tabs so you can have all of this running at the same time and because it's in Chrome it's not going to cost you memory or anything like that right it's just running in the background. So, it now basically explains what I'm wearing, who the man is. Man's wearing a watch, whatever. Short brown textured hair, subtle highlights, defined jawline, and prominent cheekbones. I like that description. It's very, it's being very like complimentary today. And then we have the clothing, everything else, right? So, that's super easy to do. So, you can use that directly here. Now, if you go back here, now we've got the video. I don't know if this is going to work, but let's test it out because I did stretch it quite far with the description, but let's see what we get back. — Why don't scientists trust Adams? Because they make up everything. — All right, so that's basically how you can animate a image. Obviously, the flying into the skyline didn't work, but I think that was just stretching it a bit too far. You can see exactly how to use this and how easy this is, right? So, let's see what else we got here. DT Workspace says, I used Gemini mobile app with Gemini camera feature in real life. No need to type in chat window to ask any question. just use camera and get the solution. Google product team solving real used problems. Yeah, 100% 100%. Actually, like I was booking a flight before and I had a picture of the passport and I had this really long form that I had to fill in. I had to grab all the individual details from the passport picture and then type it into the form. I was like, I'm not going to do that. So instead, what I actually did was I took a picture, plugged it into the app, and then said, take those detail and put it into a copied and pasteable format for the form in the order of the form. And it saved me a lot of admin, right? So it was way faster for that. So yeah, it was really good. He's got a lot of good options there to work with apps marketing brainstorm. This is really gold. Use X says what benchmarks you use for comparison of the models. Honestly, for me, like I don't trust benchmarks at all, right? Cuz as well, like it depends how many people are using the server. You don't know what kind of upgrades are running. If you look at chatt5 when I tested out the other day and you'll see this on one of my videos, I ran exactly the same prompt inside Claude as chat GPT and Claude outcoded it, right? Claude actually created an output. Chat GPT totally failed. And so the lesson there was like this AI is maybe good on the benchmarks, but in reality when you're using it dayto-day, it's actually not as good as you think. And Magic says Gemini or Manis. So it depends what you're doing. But honestly, like if you Here's the thing, right? If you go on to Manis and we'll pull this up right now. So I do man, you know, it's quite good. I actually tend to use GenSpark more, but if you use Manis and you use Gemini and we use exactly the same prompt. So we're going to say, okay, let's go back here. We'll say

yeah, we'll say inside the canvas, okay, create a quiz about SEO, right? Just something quick and easy like that. And we'll do the same inside manus exactly the same time. So we're going to ask them both side by side to create a quiz about SEO, right? So what you can see here is that Gemini, if we compare these side by side, Gemini is already coding it, right? It's already building something, which is great. Now, if you look at Manus, Manus is going to take absolutely ages cuz what it actually does is it uses browser use and computer use to go off through the internet, start trying to figure stuff out to do, then it's going to host it on an MD file, which weird, and then it's going to format it and deliver it, right? And let's have a look here. So this is the quiz. Now, if you look at this quiz, it's not interactive. It's not that interesting, etc. If you look at Gemini, look at that quiz. The quiz is way better, right? It's interactive, it's engaging, it's more fun, etc., right? And so the difference here is that Gemini tends to create better outputs at the same speed as Manis, right? Whereas Manus, it creates something basic, then you have to go back in chat, you give it another response, then you blast through a bunch more credits, and it gets a little bit annoying, right? What you can do though inside, for example, inside Manis is you can do stuff that you cannot do inside Gemini, right? For example, if we say create a 60-second video with a story about a man going to slay a dragon in the cave, it actually created all the video files here, but then it actually sliced them together and put it inside this video. So, they've got a 60-second video. You can't do that inside V3, right? And now, if we pull this up, we've got the video generated right here. This was actually an old one. So, this was a conversation from a while back, but you get the point. Like, it's quite easy to generate videos, storyboards, etc. using this process I'm showing you today. and then you can stitch everything together, right? Whereas inside Gemini, Gemini is not going to stitch stuff together for you directly, right? So that's one of the biggest differences here. But you get this like Game of Thrones feel for the content that you're creating, right? And it's really easy to do. Really easy to do step by step. Says, "If you're free, you should explain explore Amazon Bedrock. There are more than 50 LM. Yeah, I heard Amazon Bedrock is pretty good. " Actually, the new Claude update for the million tokens context window that was only available in two places. Claude Code and Amazon Bedrock. It wasn't available anywhere else. That dude says, "What LLM will allow us to schedule an agent to dedicate a task? " I can't remember if you can do scheduled tasks now inside Gemini. I think you might be able to. Yeah. So, you can launch like scheduled actions it's called. I haven't actually tested that. That is interesting. So, let's say okay, I'm going to go in here. New chat. Let's see if it's one of the options. I've never tested this out. We're going to say, okay, schedule check in AI news every day. something like that. See if it works. But yeah, for me, I mean, for scheduling, like the best thing is honestly like just a Google doc with your plan. That's what I do every day. Like I plan it out every 30 minutes. If you want to get a copy of that, then you can actually go inside the AR profit volume. Just drop me a DM and I'll get that daily planner to you. But essentially inside here, you can schedule check in different tasks here. So you can say, okay, done. I will schedule AI news for you every day. Ask me to edit or delete the scheduled thing. You get a notification when it's ready. And then if we edit this, right, we can actually edit the schedule, the instructions, AI news, and we can delete that task as well, right? So if you don't want that prompt to run, you can just delete it and then boom, you're ready to go. So that's actually pretty cool. I didn't even realize Gemini could do that, so that was really cool. Which AI system is currently the best for coding? I really love Gemini. If you're building something basic, even more complex, then I would say Claude or Gemini, but Gemini tends to have it's got three APIs for coding, so you know, you can't argue with that. DT says I find Plexi the best scheduling. Dot says coding debugging. Oh yeah. Okay. Yeah. So if you want to like debug and that sort of thing then I would recommend Claude CLI or Google CLI. You can also use chat GPT codeex. That's pretty good for coding and that sort of thing. DT recommends cl code and says is Gemini addin code assist similar to cl code. So you've got Gemini CLI right? I've got a video on this inside the AR profit boardroom if you want this, but you can use Gemini CLI and that's like an open free way to code. If you want the training on that, just go inside the AR profit boardroom, then go to Google CRI update and we walk you through exactly like how to install it, how to set it up, etc. So, it's pretty good. Very similar to CL code. Yeah. Cool. All right. So, let's start with the biggest update that honestly blew my mind when I saw it. Google just rolled out a personalization feature that lets Gemini remember your past conversations. Think about this for a second. How many times have you had to tell chat GPT or Claude the same information over and over again? You start a new chat and you're back to explaining what your business does, what your goals are, what your preferences are. Well, Google just solved that problem completely. Here's how it works. There's a new setting in Gemini that when you turn it on, it remembers key details and preferences from all your previous chats. So, if you tell it once that you run a marketing agency focused on local businesses, it remembers that forever. If you mention you prefer short bullet point answers instead of long paragraphs, it remembers that, too. This means every conversation gets smarter and more personalized. It's like having an AI assistant that actually knows you instead of meeting a

stranger every time you open a new chat. But here's the crazy part. This isn't just for paid users. This is rolling out to everyone starting with the 2. 5 Pro model. That's the free version of Gemini. Google is literally giving away features that other AI companies charge premium prices for. Now, let me show you exactly how this changes everything before this update. If I wanted help with my YouTube strategy, I'd have to explain my channel size, my niche, my goals, my target audience, everything from scratch every single time. But now, Gemini already knows I run an SEO agency. It knows I create AI content. It knows my subscriber count and engagement rates. So, when I ask for YouTube ideas, it gives me suggestions that are actually relevant to my specific situation. This is huge for business owners because now you can build a real relationship with your AI assistant. It learns your writing style, your business model, your customer base, everything. And the more you use it, the better it gets. It's like having a business partner that never forgets anything and gets smarter every day. But wait, there's more. Because Google also just launched something called temporary chat. And this solves a problem that nobody was talking about, but everyone was thinking about. Sometimes you want to ask your AI assistant something private. Maybe you're planning a surprise for someone. Maybe you're working on a confidential project. Maybe you just want to brainstorm ideas without it affecting your main conversation history. Well, now you can. Temporary chat lets you have conversations that don't get saved anywhere. Uh they don't show up in your recent chats. They don't get used to train the AI. They don't affect your personalization settings. It's like having a private conversation that disappears after 72 hours. Think about how powerful this is. You can use your main Gemini account for all your regular business stuff where you want it to remember everything, but then switch to temporary mode when you need privacy. This gives you the best of both worlds. Personalization when you want it, privacy when you need it. And here's something most people don't realize. This temporary chat feature isn't just about privacy. It's also perfect for testing different approaches without messing up your main AI relationship. Want to try a completely different writing style? Use temporary chat. Want to experiment with a new business idea without confusing your AI about your main focus? Temporary chat. Want to ask questions about a competitor without it affecting how Gemini sees your own business? You guess it, temporary chat. And speaking of getting the most out of AI, if you want to learn how to use these tools to actually make money, I've got something special for you. I've put together a complete AI training program called the AI Money Lab. The link is in the comments and description below. This covers everything from setting up these AI tools to turning them into profit generating machines for your business. Now, let's talk about the third update that honestly surprised me the most. It's called Storybook, and it's unlike anything I've seen before. Storybook lets you create complete illustrated stories from just a simple prompt. And I'm not talking about just text stories. I'm talking about stories with custom images, drawings, and illustrations that match your exact vision. Here's how wild this is. You can type something like, "Create a story about a dog who learns to code. " And Gemini will generate a full story with pictures. But here's the kicker. You can add your own photos and drawings, too. So, if you have a picture of your actual dog, you can upload it, and Gemini will create a story featuring your real dog as the main character. Your dog becomes the hero of a custom illustrated story. Now, you might be thinking, "This is just for fun, but let me blow your mind with the business applications here. " Content creators can use this to generate illustrated blog posts. Social media managers can create engaging story content with custom visuals. Authors can prototype their children's books. Educators can create custom learning materials with relevant illustrations. And here's something crazy. This works in over 45 languages. So, if you're targeting international markets, you can create story content in Spanish, French, German, whatever language your audience speaks. But the real power is in combining all three of these features together. Imagine this workflow. You use the personalization feature to teach Gemini about your business and your content style. Then you use Storybook to create illustrated content that matches your brand. And when you want to experiment with different approaches, you switch to temporary chat so it doesn't mess up your main settings. This is like having a whole content creation team built into one AI assistant. And remember, all of this is rolling out to the free version of Gemini. Google is basically giving away what will cost hundreds of dollars per month with other AI platforms. But here's what most people are going to miss. These aren't just individual features. They're building blocks for a completely new way of working with AI. The personalization means your AI gets smarter over time. The temporary chat means you can experiment without consequences. And storybook means you can create rich multimedia content without needing design skills. Put them together and you have an AI assistant that knows you personally, protects your privacy when needed, and can create professional content in minutes. Now, let me share some specific ways you can use these features starting today. For the personalization feature, start by having a conversation where you explain your business model, your goals, and your preferences. Be specific. Tell it exactly how you like information presented, what your target audience looks like, what your biggest challenges are. The more detail you give it up front, the better every future conversation will be. Think of it as training your personal AI assistant. For temporary chat, use it anytime you're working on something confidential, planning a product launch, temporary chat, researching competitors, temporary chat, testing new marketing angles, temporary chat. And for storybook, start thinking beyond just stories. This is a content creation tool disguised as a storytelling feature. You can create

illustrated tutorials, visual case studies, engaging social media content, whatever your business needs. Here's a concrete example. Let's say you run a fitness business. You could use Storybook to create an illustrated guide showing a client's transformation journey. Upload before and after photos and let Gemini create a motivational story around that transformation. Or if you're in real estate, create illustrated stories about families finding their dream homes. Use actual property photos and let Gemini weave them into compelling narratives. The possibilities are endless once you start thinking creatively. But here's the thing that excites me most about these updates. Google isn't just adding features. They're reimagining what AI assistance should look like. Most AI tools treat every conversation like you're starting from zero. But that's not how human relationships work. When you work with a real assistant, they remember what you told them yesterday. They understand your preferences. They know your business. Now AI is finally catching up to that reality. And the timing couldn't be better. We're in the middle of an AI revolution where every business needs to figure out how to use these tools effectively. The companies that master AI assistants will have a massive advantage over those that don't. These new Gemini features give you that advantage for free. Now, I want to address something important. Some people are worried about AI remembering their conversations. They think it's creepy or invasive. But here's the reality. You're in complete control. The personalization feature is optional. You can turn it off anytime. And with temporary chat, you can have private conversations whenever you want. This isn't about giving up privacy. It's about having choice. You choose when to share information and when to keep things private. And the benefits far outweigh any concerns. Having an AI that knows your business context means better suggestions, more relevant ideas, and faster results. Instead of spending time explaining the same things over and over, you can focus on getting actual work done. Here's something else that's important. These features work together to solve different problems you face throughout your day. Morning planning session where you want personalized suggestions based on your goals. Use the regular personalized mode. Midday brainstorming session about a confidential project. Switch to temporary chat. Afternoon content creation where you need illustrated materials. Fire up story book. It's like having three different AI assistants optimized for different types of work. And because it's all in one platform, you don't need to learn multiple tools or pay for multiple subscriptions. Now, let me talk about what this means for the future. Google is clearly positioning Gemini as more than just a chatbot. They're building a comprehensive AI workplace assistant. These three features are just the beginning. I expect we'll see integration with Google Workspace tools, advanced project management features, and even more sophisticated personalization options. The companies that start mastering these tools now will be way ahead when the next wave of features drops. But you don't need to wait for future updates to get value. These three features available right now can transform how you work. Start with the personalization feature. Spend 30 minutes having a detailed conversation with Gemini about your business. Tell it everything. Your industry, your customers, your goals, your challenges, your preferred communication style. Then start using it for your daily tasks and watch how much better the responses get. When you need privacy, switch to temporary chat. When you need illustrated content, use Storybook. It's that simple. Here's what happened when I put these three AI giants to the test. I gave them all the exact same challenge. Build me five different games. Same rules, same time limit, same everything. The question everyone's asking is simple. Which AI is actually better? GPT5 just dropped. Claude Opus 4. 1 is making waves. And Gemini 2. 5 Pro claims to be the smartest yet. But here's the thing. Everyone talks about these AES. Nobody actually tests them properly. So I did what nobody else would do. I put them through real challenges, not simple questions, not basic tasks. Real work that matters. What I found will change how you think about AI forever. Let me start with the first test. I asked all three A's to build a game called Pixel Ninja Dash. The goal was simple. Create a single page app where you dash and jump across rooftops while slicing enemy robots. Sounds easy, right? Wrong. Here's what happened. Gemini finished first. The UI looked good, but the game was so hard to play, I wanted to throw my computer out the window. GPT5 came second. Again, nice UI, but just like Gemini, way too hard to actually enjoy. Then Claude Opus 4. 1 finished last. Uh, but here's the twist. The game was actually fun to play. Easy controls, nice night city lights, cute background. While the other two made games that looked good but felt terrible, Claude made something you'd actually want to play. This tells us something important. Speed doesn't always win. Pretty doesn't always work. Sometimes the AI that takes longer makes something better. But wait, it gets more interesting. Test two was candy match blast. Match three candies. Score points. Beat the clock. Standard match three game that everyone knows. Gemini finished first again. But there was an error. The game didn't work. So while I waited for Gemini to fix it, Claude finished. And holy cow, instead of regular candy graphics, Claude used emojis, chocolates, candies, sweets. It looked completely different from what the others made. GPT5 finished next. The

design was like painted balls. Pretty and polished. It worked well. Finally, Gemini fixed its error and came last. When it finally worked, I wasn't impressed. The design was boring compared to the others. Claude won this round. Not because it was fastest, because it was creative. It thought outside the box. It made something unique while the others played it safe. This is where most people get AI wrong. They think faster means better. They think first place means best. But what actually matters is results. What matters is whether the thing you ask for actually works the way you need it to. Test three was jungle run adventure. Run and jump across jungle platforms. Collect bananas. Avoid traps. Infinite terrain that gets faster as you play. Here's where things got messy. All three AIs failed on the first try. Every single one had errors. So, I gave them all a second chance. Claude finished first this time. The game design looked good, but there were no bananas, no collectibles, just a monkey running. I gave it another chance. The second version didn't work at all. Gemini came next. It worked most of the time, but sometimes the space bar didn't respond. Sometimes you couldn't jump. And instead of bananas, it made yellow balls. I'm not kidding. Yellow balls instead of bananas. GPT5 came last. The game crashed completely three times. It just couldn't handle this challenge at all. This test showed me something crucial. When AIs fail, they fail differently. Some fail fast and recover. Some fail slow but work eventually. Some just keep failing no matter what you do. But here's what's really happening behind the scenes. Each AI has different strengths, different ways of thinking about problems, different approaches to code. And here's where you need to pay attention because the next test changed everything. Test four was Space Miner 3D. Fly a spaceship through asteroids. Collect crystals. Don't crash. Use 3. js JS for 3D graphics. Gemini finished first. Nothing special. Standard game. GPT5 came second. Also nothing special. Same basic approach. But Claude came last. At first it showed errors. I thought it failed again. But then I realized something. The game was actually working and it looked way better than the other two. Despite the initial errors, Claude had built something more impressive. This taught me the biggest lesson of all. Don't judge an AI by its first response. Don't judge it by how fast it works. Judge it by the final result. You know what? Most business owners get wrong about AI. They pick the AI that gives them answers fastest. They want quick results, fast responses, instant gratification. But that's like choosing a restaurant because they serve food fast. Speed doesn't mean quality. Speed doesn't mean the food tastes good. Speed doesn't mean you'll be satisfied. The same is true with AI. The best AI isn't always the fastest AI. Now, here's where you can transform your business with AI. Most people use AI wrong. They ask simple questions. They want quick answers. They treat AI like a search engine. That's like buying a Ferrari and only driving it to the grocery store. The real power of AI comes when you push it. When you give it complex challenges, when you test its limits. And that's exactly what the AI success lab teaches you. How to use AI the right way. How to get results that actually matter. How to save hundreds of hours and thousands of dollars with the right AI strategies. Check the link in the comments and description. Over 14,000 members are already inside learning the advanced techniques that most people never discover. But let me finish the story first because test 5 was the real eye opener. Test 5 was Lava Escape Runner. Run across crumbling platforms while lava rises from below. Volcanic theme, fiery colors, intense but not frustrating gameplay. Gemini finished first again, but guess what? Error again. So I gave it time to fix it. Claude came second. The game worked. Sometimes the controls didn't respond perfectly, but it was playable. It was fun. GPT5 came last, and this is where it completely fell apart. Error after error, three times in a row, it just couldn't handle this challenge. So, what does all this mean? Which AI actually wins? Here's the truth. There is no single winner. Each AI has different strengths. Gemini is fast, but makes more errors. GPT5 is polished, but struggles with complex tasks. Claude is creative and often builds better final products, even if it takes longer. But here's what really matters for your business. You need to know which AI to use for which task. For quick, simple tasks, Gemini might be your best bet. For polished standard work, GPT5 could work well. For creative, complex projects, Claude often surprises you with something better. Most people don't know this. They pick one AI and stick with it. That's like using a hammer for every job. Sometimes you need a screwdriver. saw. The businesses that are winning with AI right now understand this. They use different AIs for different purposes. They know when to use which tool. And that's exactly what we teach inside the AI success lab. Not just how to use one AI, but how to use all of them. How to pick the right tool for each job. How to get results that your competitors can't match. You see how I show a checklist of 100 different tutorials that were given

away as freebies every day inside the school feed. That's just the beginning. You get all the video notes, all the strategies, all the techniques that took me years to figure out. Plus, you join 14,000 other members who are all learning and sharing the latest AI breakthroughs. When you're part of something bigger than yourself, you learn faster. You grow faster. You get results faster. Here's what this really means for you. AI isn't just about getting quick answers anymore. It's about building systems, creating processes, automating tasks that used to take hours. The companies that figure this out first will dominate their markets. The ones that don't will be left behind. I've seen it happen already. Businesses that used to spend thousands on content creation now do it for pennies. Companies that used to hire expensive developers now build apps with AI. Entrepreneurs who used to struggle with marketing now create campaigns that convert like crazy, but only if they know how to use AI properly. I decided to run the ultimate test. Four of the most powerful AI models on the planet all fighting for one thing. The title of best AI coder. We've got Quen 3-235B- A22B-2507. This is Alibaba's brand new monster. Just dropped and it's already making waves. This thing is their latest coding powerhouse that's designed to compete with the best. Then we have Claude for Opus Anthropics coding beast. They claim it's the world's best coding model. It can code for 7 hours straight without stopping. That's like having a developer who never gets tired. Next up is Gemini 2. 5 Pro, Google's latest and greatest. This model is supposed to be revolutionary. Google says it can handle any coding challenge you throw at it. And finally, Gro 4. This is the wild card. It's got that X factor that might surprise everyone. Here's the thing, though. Everyone talks about which AI is best, but nobody runs real tests. They just look at benchmarks. But benchmarks don't tell the whole story. You need real world challenges. So I created seven brutal tests. These aren't simple coding tasks. These are the kind of projects that separate the winners from the losers. Let me tell you about test number one. HTML game with falling objects clicker. Here's the exact prompt I gave each AI. Build a simple HTML and JavaScript game where objects fall from the top of the screen and the player has to click them before they hit the ground. The game should include a start screen, score tracking, increasing difficulty over time, and work well on both desktop and mobile. I asked each AI to build a complete game, not just some basic code, a real game with all these features. This is where things got interesting. Gemini came out swinging, super fast response. The game actually worked. You could play it right away. No bugs, no crashes, just pure gaming goodness. Quen 3 coder was next. Built almost the same game. Pretty solid stuff. The mechanics worked great. Grock 4 followed with similar output. Nothing groundbreaking, but it did the job. But here's the twist. Claude 4 Opus was last to finish, but when I saw the UI, my jaw dropped. It was completely different from the others. Better design, cleaner interface, more professional looking. So, round one goes to Gemini for speed and functionality. But Claude gets style points. Test number two was brutal. A it tool landing page. The prompt was, "Create a clean, modern landing page using only HTML and CSS for a fictional AI tool called Prompt Pilot. It should include a catchy headline, sub headline, hero, image, placeholder, a section with three features using icons, a testimonial carousel, and a bold animated call to action button. I wanted a complete landing page for this fake AI tool. Had to be modern, clean with all these specific elements. Gemini strikes again, first to finish, the landing page looks amazing. Great font combinations, professional design. This thing could fool anyone into thinking it was a real product. Quen came next, but the page was super plain. The text color was so light you could barely see it. Major fail on the user experience front. Grock 4 delivered something basic, not impressive. Looked like a landing page from 2015. Claude 4 Opus came last again, but wow, the UI was gorgeous. Perfect color choices, professional layout. If I was running a real AI company, I'd use this design. Here's what I'm seeing. Gemini is fast and functional. Claude is slow but beautiful. Quen 3-235B- A22B-2507 is hit or miss. Grock is basic. Test three changed everything. Multiplayer tic-tac-toe in the browser. Here's the prompt I used. Build a two-player tic-tac-toe game using HTML, CSS, and JavaScript that works in real time via websockets or a similar method. It should allow two users to play from different browsers with automatic wind detection and a restart option. This wasn't just any tic-tac-toe game. This had to work in real time. Two players in different browsers, websocket connections, wind detection, restart options. This is advanced stuff. Plot twist. Quen 3-235B- A 22B-2507

finish first. And the game actually worked. Nice UI, smooth gameplay, everything functioned perfectly. Claude came second, good UI, working game, solid performance. Grock 4 was supposed to deliver a game, but when I checked, there was no game. Complete failure. Gemini came last with a working game, but the UI looked terrible. Functional, but ugly. This is where Quen 3-235B- A22B-2507 showed its true power. When it comes to complex interactive applications, it's a beast. Test four was the browserbased markdown editor. The prompt make a browserbased markdown editor using HTML, CSS, and JavaScript. The page should have a split screen layout with a markdown input on the left and a live preview on the right. Include buttons for bold, italic, and links, and allow the user to export the content as a MD file. Split screen layout, live preview, export functionality. This is developer tool territory. Quen finish first. But here's the problem. When I tried to preview anything, it threw errors. I tried to fix it, but the errors kept coming. Total disaster. Gemini came next with a working Markdown editor. Clean, functional, did exactly what it was supposed to do. Grock and Claude both built editors. But when you tried to pace basic HTML, neither would preview it properly. Both failed the real world test. So Gemini takes this round for actually delivering something that works. Here's where things get crazy. Test five was portfolio website generator. My prompt was build a simple web tool where a user can enter their name, skills, and project descriptions. And when they click a generate button, it creates a styled onepage portfolio website, include an option to download the generated HTML file. I wanted a tool where users could enter their info, click generate, and boom, a complete portfolio website appears, plus the ability to download the HTML file. Jim and I finished first with a nice UI, but it had a fatal flaw. It didn't respond to user input. Beautiful, but broken. Quen had errors again. Starting to see a pattern here. Claude delivered something special. Good UI, working buttons, responsive design. Actually did the job it was supposed to do. Grock had the same functionality as Claude, but the UI looked basic. Still worked though. Claude wins this round for delivering both form and function. Test six blew my mind. Tower defense game. Here's the prompt. Create a simple 2D tower defense game using HTML 5 canvas and JavaScript where enemies follow a path and the player places towers to stop them. Include waves of increasing difficulty, basic tower upgrades, and a game over screen. This isn't just a game. This is a complete 2D tower defense system. Enemies following paths, towers that shoot, waves of increasing difficulty, tower upgrades, game over screens. This is serious game development. Quen finished first, but I couldn't figure out how to play the game. The UI was confusing. The mechanics weren't clear. It technically worked, but was unusable. Then Gemini delivered something incredible. This felt like a real game. Professional fonts, smooth mechanics, perfect UI. It was like playing a game from the app store. Absolutely insane for just one prompt. Gro's game didn't work properly. Major disappointment. Claw did the job, but it could use improvement. Functional, but not impressive. Gemini completely destroyed the competition here. No contest. Test seven was the ultimate challenge. First person maze escape 3D game. The final prompt. Create a basic firstperson 3D maze escape game in the browser using 3JS. The player should be able to move through a 3D maze using keyboard controls WD. Find a key and unlock an exit door to win. I wanted a full 3D maze game using 3JS. First person perspective keyboard controls with WD. Find a key, unlock an exit door. win the game. This is professional game development territory. Gemini was incredible. Created a full 3D maze that actually worked. You could move around, find the key, escape through the door. It felt like playing a real video game. Quen showed just a brown background. No interaction. Complete failure. Claude impressed me. Built a working 3D maze. You could navigate through it, find objectives, escape to win. Really solid implementation. Grock just showed a white background. Another complete failure. Gemini wins again with the most impressive 3D game. So, let's talk about what this all means. After running seven brutal tests, here's what I discovered. Gemini 2. 5 Pro is the speed demon. Fast responses, great UI design. When it works, it really works. Especially impressive with games and visual applications. Claude for Opus is the perfectionist. takes longer to respond but delivers higher quality code, better UI designs, more professionallook results. When you need something that looks good and works well, Claude delivers. Quen 3-235B- A22B-2507 is the wild card. Sometimes it's brilliant. The multiplayer tic-tac-toe was amazing, but sometimes it fails completely. Inconsistent, but has

moments of genius. Grock 4 is the underdog. Basic, but functional most of the time. Not exciting, but gets the job done for simple projects. Here's the thing, though. The winner depends on what you need. If you want speed and impressive visual results, Gemini is your choice. It's perfect for prototypes and demos that need to wow people. If you want professional quality code that looks good and works reliably, Claude for Opus is the winner. Yes, it's slower, but the results are worth waiting for. If you're feeling adventurous and want to try something with huge potential, Quen 3-235B- A22B-2507 might surprise you. Just be prepared for some inconsistency. And if you need something basic that just works, Gro 4 will do the job. But here's the real secret. The best coders don't rely on just one AI. They use multiple models for different tasks. Use Gemini for rapid prototyping and visual applications. Use Claude for production code that needs to be professional. Use Quen 3-235B- A22B-2507 when you want to experiment with cuttingedge capabilities. I decided to run the ultimate test. Four of the most powerful AI models on the planet. All fighting for one thing, the title of best AI coder. We've got Quen 3-2 35B- A22B-2507. This is Alibaba's brand new monster just dropped and it's already making waves. This thing is their latest coding powerhouse that's designed to compete with the best. Then we have Claude for Opus, Anthropics coding beast. They claim it's the world's best coding model. It can code for 7 hours straight without stopping. That's like having a developer who never gets tired. Next up is Gemini 2. 5 Pro, Google's latest and greatest. This model is supposed to be revolutionary. Google says it can handle any coding challenge you throw at it. And finally, Gro 4. This is the wild card. It's got that X factor that might surprise everyone. Everyone talks about which AI is best, but nobody runs real tests. They just look at benchmarks, but benchmarks don't tell the whole story. You need real world challenges. So, I created seven brutal tests. These aren't simple coding tasks. These are the kind of projects that separate the winners from the losers. Let me tell you about test number one. HTML game with falling objects clicker. Here's the exact prompt I gave each AI. Build a simple HTML and JavaScript game where objects fall from the top of the screen and the player has to click them before they hit the ground. The game should include a start screen, score tracking, increasing difficulty over time, and work well on both desktop and mobile. I asked each AI to build a complete game, not just some basic code. A real game with all these features. This is where things got interesting. Gemini came out swinging, super fast response. The game actually worked. You could play it right away. No bugs, no crashes, just pure gaming goodness. Quen 3 coder was next. Built almost the same game. Pretty solid stuff. The mechanics worked great. Grock 4 followed with similar output. Nothing groundbreaking, but it did the job. But here's the twist. Claude 4 Opus was last to finish, but when I saw the UI, my jaw dropped. It was completely different from the others. Better design, cleaner interface, more professional looking. So, round one goes to Gemini for speed and functionality, but Claude gets style points. Test number two was brutal, a it eye tool landing page. The prompt was create a clean modern landing page using only HTML and CSS for a fictional AI tool called prompt pilot. It should include a catchy headline, sub headline, hero, image placeholder, a section with three features using icons, a testimonial carousel, and a bold animated call to action button. I wanted a complete landing page for this fake AI tool. Had to be modern, clean with all these specific elements. Gemini strikes again, first to finish. The landing page looked amazing. Great font combinations, professional design. This thing could fool anyone into thinking it was a real product. Quen came next, but the page was super plain. The text color was so light you could barely see it. Major fail on the user experience front. Grock 4 delivered something basic, not impressive. Looked like a landing page from 2015. Claude 4 Opus came last again, but wow, the UI was gorgeous. Perfect color choices, professional layout. If I was running a real AI company, I'd use this design. Here's what I'm seeing. Gemini is fast and functional. Claude is slow but beautiful. Quen 3-235B- A22B-2507 is hit or miss. Grock is basic. Test three changed everything. Multiplayer tic-tac-toe in the browser. Here's the prompt I used. Build a two-player tic-tac-toe game using HTML, CSS, and JavaScript that works in real time via websockets or a similar method. It should allow two users to play from different browsers with automatic wind detection and a restart option. This wasn't just any tic-tac-toe game. This had to work in real time. Two players in

different browsers, websocket connections, wind detection, restart options. This is advanced stuff. Plot twist. Quen 3-235B- A22B-2507 finished first and the game actually worked. Nice UI, smooth gameplay, everything functioned perfectly. Claude came second, good UI, working game, solid performance. Grock 4 was supposed to deliver a game, but when I checked, there was no game. Complete failure. Gemini came last with a working game, but the UI looked terrible. Functional, but ugly. This is where Quen 3-235B- A22B-2507 showed its true power. When it comes to complex interactive applications, it's a beast. Test 4 was the browserbased markdown editor. The prompt, make a browserbased markdown editor using HTML, CSS, and JavaScript. The page should have a split screen layout with a markdown input on the left and a live preview on the right. include buttons for bold, itallic, and links and allow the user to export the content as a MD file. Split screen layout live preview export functionality. This is developer tool territory. Quen finish first. But here's the problem. When I tried to preview anything, it threw errors. I tried to fix it, but the errors kept coming. Total disaster. Gemini came next with a working Markdown editor. Clean, functional, did exactly what it was supposed to do. Grock and Claude both built editors, but when you tried to paste basic HTML, neither would preview it properly. Both failed the real world test. So, Gemini takes this round for actually delivering something that works. Here's where things get crazy. Test five was portfolio website generator. My prompt was build a simple web tool where a user can enter their name, skills, and project descriptions. And when they click a generate button, it creates a styled onepage portfolio website. Include an option to download the generated HTML file. I wanted a tool where users could enter their info, click generate, and boom, a complete portfolio website appears, plus the ability to download the HTML file. Jim and I finished first with a nice UI, but it had a fatal flaw. It didn't respond to user input. Beautiful, but broken. Quen had errors again, starting to see a pattern here. Claude delivered something special. Good UI, working, responsive design. Actually did the job it was supposed to do. Grock had the same functionality as Claude, but the UI looked basic. Still worked, though. Claude wins this round for delivering both form and function. Test six blew my mind. Tower defense game. Here's the prompt. Create a simple 2D tower defense game using HTML 5 canvas and JavaScript where enemies follow a path and the player places towers to stop them. Include waves of increasing difficulty, basic tower upgrades, and a game over screen. This isn't just a game. This is a complete 2D tower defense system. Enemies following paths, towers that shoot, waves of increasing difficulty, tower upgrades, game over screens. This is serious game development. Quen finished first, but I couldn't figure out how to play the game. The UI was confusing. The mechanics weren't clear. It technically worked, but was unusable. Then Gemini delivered something incredible. This felt like a real game. Professional fonts, smooth mechanics, perfect UI. It was like playing a game from the App Store. Absolutely insane for just one prompt. Grock's game didn't work properly. Major disappointment. Claw did the job, but it could use improvement. Functional, but not impressive. Gemini completely destroyed the competition here. No contest. Test 7 was the ultimate challenge. First person maze escape 3D game. The final prompt, create a basic firstperson 3D maze escape game in the browser using 3JS. The player should be able to move through a 3D maze using keyboard controls WD. Find a key and unlock an exit door to win. I wanted a full 3D maze game using 3JS. First person perspective keyboard controls with WD. Find a key, unlock an exit door, win the game. This is professional game development territory. Gemini was incredible. Created a full 3D maze that actually worked. You could move around, find the key, escape through the door. It felt like playing a real video game. Quen showed just a brown background. No interaction. Complete failure. Claude impressed me. Built a working 3D maze. You could navigate through it, find objectives, escape to win. Really solid implementation. Grock just showed a white background. Another complete failure. Gemini wins again with the most impressive 3D game. So, let's talk about what this all means. After running seven brutal tests, here's what I discovered. Gemini 2. 5 Pro is the speed demon. Fast responses, great UI design. When it works, it really works. Especially impressive with games and visual applications. Claude for Opus is the perfectionist. Takes longer to respond but delivers higher quality code, better UI designs, more professionallook results. When you need something that looks good and works well, Claude delivers. Quen 3-235B-

A22B-2507 is the wild card. Sometimes it's brilliant. The multiplayer tic-tac-toe was amazing, but sometimes it fails completely. Inconsistent, but has moments of genius, Grock 4 is the underdog. Basic, but functional most of the time. Not exciting, but gets the job done for simple projects. Here's the thing, though. The winner depends on what you need. If you want speed and impressive visual results, Gemini is your choice. It's perfect for prototypes and demos that need to wow people. If you want professional quality code that looks good and works reliably, Claude for Opus is the winner. Yes, it's slower, but the results are worth waiting for. If you're feeling adventurous and want to try something with huge potential, Quen 3-2 35B- A22B-2507 might surprise you. Just be prepared for some inconsistency. And if you need something basic that just works, Grock 4 will do the job. But here's the real secret. The best coders don't rely on just one AI. They use multiple models for different tasks. Use Gemini for rapid prototyping and visual applications. Use Claude for production code that needs to be professional. Use Quen 3-235B- A22B-2507 when you want to experiment with cuttingedge capabilities. Everyone's talking about which AI is best, but nobody's actually testing them on real business tasks. Everyone just asks them to write a poem or solve a math problem. and that's useless. So, I decided to do something different. I created five brutal business tests, real tasks that you and I face every single day. SEO analysis, content creation, strategic planning, the works, and I put three of the hottest AI models through the ringer, Grock 4, Claude for Sonnet, and Gemini 2. 5 Pro. What I found will change how you think about AI forever. Let me break this down for you. First, let me tell you about the testing method because this wasn't some casual comparison. This was scientific. This was real. I created five different business scenarios. Each one designed to test a different skill that matters in business. Researchability, content creation, strategic thinking, technical knowledge, speed. Every test had clear evaluation criteria. I measured speed. I measured accuracy. I measured quality. And I scored everything on a point system so there was no bias. And I'm going to share the exact prompts I used. So you can test these yourself. You can see if you get the same results. You can use these prompts in your own business. Test one was SEO competitive analysis. This is something I do every single day in my agency. I gave each AI model a scenario where they're running a project management software company. They had to analyze competitors like Asana and Monday. com, find keyword opportunities, identify content gaps, create a six-month strategy. Here's the exact prompt I used. I run a midsize SAS company selling project management software. Our main competitors are Asana, Monday. com, and Trello. Create a comprehensive SEO competitive analysis that includes one, identify 10 highv value keyword opportunities where we could outrank these competitors. Two, analyze their content gaps we could exploit. Three, recommend a six-month SEO strategy with specific tactics and expected timeline for results. Include reasoning for your recommendations. This is complex stuff. This requires research skills, strategic thinking, business understanding. And here's what happened. Grock 4 absolutely crushed this test. It delivered the most comprehensive report I've ever seen from an AI. Detailed tables, extensive data analysis, strategic insights that actually made sense. The organization was perfect. The visual presentation was clean. Claude the fourth Sonic came in second. Strong analytical depth, good strategic recommendations. well structured but just not as comprehensive as Grock. Gemini 2. 5 Pro. It was the fastest by far, but the analysis was basic, surface level, and not enough depth for real strategic decision-making. So, after test one, Grock was leading. Test two was content creation. I asked each AI to write a complete 1,500word blog post for uh cyber security company. Target keyword was endpoint security best practices 2025. had to include SEO optimization, proper heading structure, actionable advice, internal linking opportunities, a compelling call to action. Here's the exact prompt. Write a complete 1,500word blog post for our B2B cyber security company targeting IT directors at midsize companies with 500 to 2,000 employees. Primary keyword endpoint security best practices 2025 include one SEO optimized title and meta description. Two, proper heading structure H1 to H3. Three, five actionable best practices with real world examples. Four, internal linking opportunities suggest. Three, five, a compelling CTA for our endpoint security audit. Make it authoritative but

accessible avoiding technical jargon. This is bread and butter content marketing something every business needs. and Claude Forson on it destroyed this test. It delivered 1,620 words of absolutely beautiful content. The SEO optimization was perfect. Natural keyword integration. The structure was clean. The readability was incredible. It felt like a human expert wrote it. Gemini 2. 5 Pro came in second with 1,529 words. Good content quality, proper SEO elements, solid performance across the board. But here's where things got interesting. Gro 4 had a massive problem. It claimed it wrote 1,512 words, but when I counted only 1,221 words. That's a huge accuracy issue. You can't trust an AI that lies about word count. What else is it lying about? So, after test two, Claude was catching up. Test three was content repurposing. I gave them a high performing blog post about remote team management. 3,000 words that generated 15,000 organic views and 200 leads. Real numbers from real campaigns. I asked them to turn this into eight different content formats. LinkedIn carousel, email newsletter series, Twitter thread, YouTube script, infographic outline, podcast brief, Instagram stories, short form video script. Here's the exact prompt. We have a high-erforming 30,000word blog post about remote team management strategies that generated 15,000 organic views and 200 leads. Transform this single piece into eight different content formats for maximum reach. One, LinkedIn carousel post. Two, email newsletter series. Three-part. Three, Twitter thread. Four, YouTube video script 8 to 10 minutes. Five, infographic outline. Six, podcast episode brief. Seven, Instagram story series. Eight short form video script 60 seconds. Provide the actual content scripts, not just ideas. This tests creativity adaptation skills, platform understanding. Claude for Sonic won again, and it wasn't even close. The copyrightiting was exceptional. The hooks were incredible. Each piece of content was perfectly optimized for its platform. The quality stayed high across all eight formats. Gemini 2. 5 Pro came second. Good adaptation skills, solid content, but missing that creative spark. Grock 4 was adequate, functional, but boring. No compelling hooks, no engagement. Claude was pulling ahead. Test four was editorial calendar development. Create a 90-day content calendar for a project management tool. Target keywords around project management tips and team productivity. Include content clusters, seasonal topics, lead magnets, promotion strategy. Here's the exact prompt I used. Create a 90-day content calendar for our SAS project management tool targeting small business owners and team leaders. We want to rank for project management tips, team productivity, and small business efficiency. Include one, 24 blog post topics with target keywords and search intent. Two, content cluster strategy linking related posts. Three, seasonal trending topic integration. Four, lead magnet content three pieces. Five, content promotion strategy for each post format as a detailed calendar with publication dates, word counts, and success metrics. This is strategic planning at its finest. And here's where things got interesting. Both Grock 4 and Gemini 2. 5 Pro tied for first place. Both delivered excellent table-based organization, comprehensive planning, detailed timelines and metrics, strategic thinking, visual clarity that made the plans immediately actionable. Claude the fourth sonic came third. good analysis and recommendations, comprehensive, but it lacked the superior organization of the other two. So, we had a three-way race going into the final test. Test five was technical SEO audit. I gave them a 500page e-commerce website with multiple technical issues, duplicate title tags, slow site speed, broken links, missing structured data, missing alt text, outdated XML sitemap, limited development resources, only 20 hours per month. I'm the CMO of a 500page e-commerce website. Our developer found these technical issues. One, 127 pages with duplicate title tags. Two, site speed averaging 4. 2 seconds on mobile. Three, 23 broken internal links. Four, missing structured data on product pages. Five, 15% of images missing alt text. Six, XML sitemap hasn't been updated in 8 months. We have limited development resources 20 hours per month. Create a priority matrix ranking these issues by SEO impact versus implementation effort and provide a three-month implementation road map. They had to create a priority matrix rank issues by SEO impact versus implementation effort. Provide a three-month road map. Both Grock 4 and Claude for Sonnet tied for first. Excellent strategic road maps, strong prioritization logic, clear implementation paths, deep technical SEO

understanding, practical resource allocation. Gemini 2. 5 Pro came third. Adequate technical understanding, but missing the strategic depth and implementation detail. But wait, I wasn't done testing yet. I wanted to see how these AIs handled something completely different, game development, because if you can build games, you can build anything. So, I threw two more brutal tests at them. Test six was building a Minecraft inspired 3D world generator using p5. gs. This is hardcore coding, real programming skills. Here's the exact prompt. Build a Minecraft inspired 3D world generator using p5gs. Create one, a 20x20 grid of procedurally generated terrain with different block types, grass, stone, dirt, water. Two, firsterson camera controls. Where's the movement mouse look? Three, block placement removal system, left right click. Four, at least four different block textures. Five, simple lighting or shading effects. Six, inventory system showing available blocks. Seven, save, load, world state functionality. Focus on performance. It should run smoothly at 30 plus frames pers. Include chunkbased loading for scalability. This is insanely complex. 3D graphics, game mechanics, performance optimization, user interface, data persistence. Claude for Sonet won this round. It actually created a game not fully playable, but it built something that resembled what I asked for. The code structure was solid. The approach was logical. Grock 4 and Gemini 2. 5 Pro, they finished fast, but the code was full of errors, broken functionality, incomplete features, speed without substance. Then I hit them with test seven, a 2D endless runner game. Here's the prompt. Create a complete 2D endless runner game using HTML 5 canvas and JavaScript. The game should include one, a character that can jump to avoid obstacles. Two, randomly generated obstacles moving from right to left. Three, score system that increases over time. Four, game over screen with restart functionality. Five, progressive difficulty. Obstacles get faster. Six, simple particle effects when jumping or colliding. make it fully playable with smooth animations and responsive controls. Include proper game state management and clean commented code. Clawed Force on it absolutely destroyed this test. It created a completely playable game. Smooth animations, responsive controls, progressive difficulty, particle effects, game state management, everything worked perfectly. The game was actually hard but so fun to play like a real indie game you'd find on Steam. 5 Pro. Same story. Fast responses, broken code, lots of errors, incomplete functionality. This revealed something huge about these AIs. So, what were the final results after all seven tests? Claude for Sonic won four solo victories and one co-win. Six out of seven strong performances. Absolutely dominant. Grock 4 won one solo victory and two co-wins. Three out of seven strong performances. Gemini 2. 5 Pro won zero solo victories and one co-ind. One out of seven strong performances. But here's what really matters. The insights. Claude for Sonnet isn't just good at content creation. It's exceptional at complex coding tasks. Game development, interactive applications, advanced programming challenges. When you need something that actually works, Claude delivers. Gro 4 and Gemini 2. 5 Pro are fast. But fast doesn't matter if the code is broken, if the functionality doesn't work, if you can't actually use what they create. This is the difference between an AI that can help you build a real business versus an AI that just gives you pretty responses that don't work. Claude Forson is the king of content creation, blog posts, copywriting, content repurposing. If you're doing content marketing, this is your AI. The quality is consistently excellent. The engagement is superior. The readability is perfect. But that's not all. Claude also dominates complex coding tasks, game development, interactive applications, advanced programming. When you need code that actually works, Claude is your AI. Grock 4 is the analysis beast. Competitive research, strategic planning, comprehensive reporting. If you need deep analysis, and strategic thinking, this is your AI. The organization is excellent. The visual presentation is clean. Gemini 2. 5 Pro is the speed demon. Fastest response times by far. Good content quality for basic tasks. Reliable performance for simple requests. If you need quick turnaround, and speed is more important than quality or functionality, this is your AI. But here's the thing that nobody talks about. Different A is for different tasks. You don't need to pick just one. In my agency, we use Claude for all our content creation, blog posts for clients, email campaigns, social media content. It's unmatched. We use Grock for competitive analysis, market research, strategic planning. When we need comprehensive reports for clients, we use Gemini when we need quick answers, fast research, rapid content for social media, time-sensitive projects. It's like having different

tools in a toolbox. You don't use a hammer for everything. But if I had to pick just one AI for business, Claude Force on it, here's why. Content is everything in business today. Content marketing drives leads, copywriting drives sales, email campaigns drive revenue, social media content builds brands, and Claude dominates content creation. But here's what shocked me the most. Claude also dominates coding, real development work, functional applications, interactive tools. This means Claude can help you build the entire business ecosystem. Content to attract customers, code to serve customers, everything you need in one AI. Plus, Claude's consistency is incredible. Every single test, it delivered highquality results. You can trust it. Grock's word count issue in the blog post test is concerning. If it lies about something basic like word count, what else might it get wrong? Speed is nice, but quality matters more. Would you rather have a fast AI that gives you mediocre content or a slightly slower AI that gives you incredible content? I know which one I choose, but only if you use it correctly. Only if you go beyond the surface level stuff, only if you treat it like the powerful business tool it actually is. These test results prove that AI can handle complex business tasks, real work, real value. The question is, are you ready to take advantage of it? use AI to scale your business, get more customers, save hundreds of hours, or are you going to keep using it to write poems? The choice is yours. But I know what successful entrepreneurs choose. They choose results over entertainment, business value over party tricks, real applications over fun experiments. That's what separates successful people from everyone else. They see the opportunity, they take action, they get results. If you want to join them, scale your business with AI, I have three ways I can help you right now. First, if you're serious about using AI to scale your business, check out my AI profit boardroom. It's the best place to scale your business, get more customers, and save hundreds of hours with AI automation. We focus on real business applications that generate real revenue. Second, if you want help with the SEO and getting more organic traffic, book a free SEO strategy session. The link is in the comments and description. My team will analyze your website and show you exactly how to get more leads and customers from Google. Third, if you want to master AI for business, check out the AI success lab. We have SOPs and processes for over 100 different business use cases, plus video tutorials that walk you through everything step by step. You can see we have over 14,000 members because people don't want to miss out. They want to be part of something bigger. And you can get all the video notes from today's analysis, plus access to our entire community of entrepreneurs who are scaling with AI. Julian Goldie reads every comment, so make sure you comment below and let me know which AI you're going to start using for your business. Thanks for watching and I'll see you in the next video. Google just dropped Gemini 2. 5 flashlight and it's their fastest, cheapest AI model ever. But here's the crazy part. I spent hours testing this thing and the results shocked me. We're talking about an AI that costs almost nothing but beats expensive models at coding games. I built multiple games in seconds. I compared it head-to-head with premium models and what I found will blow your mind. This could change everything about how we use AI. Hey, if we haven't met already, I'm the digital avatar of Julian Goldie, CEO of SEO agency Goldie Agency. Whilst he's helping clients get more leads and customers, I'm here to help you get the latest AI updates. And today, we're talking about something massive. Google just announced Gemini 2. 5 Flashlight. And when I say just announced, I mean like hours ago. Now, Google's calling this their fastest and cheapest 2. 5 model yet. But here's what got my attention. They're claiming it beats their old models on everything. Coding, math, science, reasoning, everything. So, I did what I always do. I ignored the marketing hype and I tested it myself for hours. Real world tests, no BS benchmarks that don't mean anything. I built games. I created landing pages. I compared it directly with Gemini 2. 5 Pro, which costs way more money. And what I found was shocking. Because here's the thing, this model costs almost nothing to use. We're talking fractions of a cent per request. But in some tests, it actually beat the expensive models. That's not supposed to happen. So, in this video, I'm going to show you exactly what Gemini 2. 5 flashlight can do. The real results, not the marketing fluff. I'll show you where it crushes the competition and where it falls flat on its face. Plus, I'll show you exactly how to use it for free right now. No credit card needed because if this model is as good as Google claims, it could save you hundreds or even thousands of dollars on AI costs while actually giving you better results. But first, let me explain what this thing actually is and why it matters. So, here's what Google just dropped on us. Gemini 2. 5 flashlight is their new

preview model. And when they say preview, they mean you can use it right now for free. No credit card, no payment required. Just go to AI Studio and start using it. Now, Google's calling this their most costefficient and fastest 2. 5 model yet. But here's what makes it different from everything else they've released. It comes in two modes: thinking mode and non-thinking mode. In thinking mode, the AI actually thinks through problems before answering. Like you can literally see it working through the problem step by step. This gives you better results, but takes a bit longer and costs more if you're using the API. In non-thinking mode, it just gives you answers fast. Super fast. Like crazy fast. And here's the kicker. Even in non-thinking mode, it still beats the old 2. 0 flashlight on everything. Coding, math, science, reasoning, everything. But the real question is, how does it perform in the real world? Because benchmarks are nice and all, but I care about one thing. Can this thing actually help me build stuff that works? So, I put it through my brutal real world test. The same tests I use for every AI model. I started with something simple. I asked it to build a Space Invaders game. Complete HTML, CSS, JavaScript, one prompt, make it look amazing, and trigger loads of dopamine. Now, here's where it gets interesting. I ran the exact same prompt on Gemini 2. 5 Pro at the same time. Remember, 2. 5 Pro costs way more money. Is supposed to be the smart, powerful model. Flashlight finished first. Way first. But that's not the crazy part. The crazy part is Flashlight actually worked perfectly on the first try. The game looked good. It played smooth. Everything worked. Gemini 2. 5 Pro, it failed. Completely failed the oneshot test. didn't even create a working game. I'm staring at my screen thinking, "Wait, what? The cheap, fast model just beat the expensive smart model. But maybe that was just luck. " So, I tested it again. This time, I asked both models to build a dopamine drums app, something visual with sound effects. Again, one prompt. Flashlight coded it out super fast, added all the visual effects. But here's the thing, it forgot the audio. So, I just told it add audio and boom, fixed instantly. Gemini 2. 5 Pro. Still coding, taking forever. And when it finally finished, the output was worse. Way worse. Now I'm starting to think something's going on here. So I pushed it harder. I asked it to build a Sonic game as a single HTML file. This is where most AI models start to break down. Flashlight delivered a playable Sonic game. Not perfect, but playable. The graphics were basic, but it worked. Then I asked it to build a Flappy Bird game. Again, one shot. And you know what? It nailed it. The game worked perfectly. The physics felt right. The difficulty was spoton. I'm playing this Flappy Bird game built by a model that costs almost nothing. And I'm thinking, "This changes everything. " But wait, there's more. Remember how I said this thing is fast? Let me show you exactly how fast. I used it to build a complete SEO calculator landing page, beautiful design, working JavaScript, responsive layout, the whole thing. It took seconds, not minutes, seconds. I've used Claude Sonic 4, GPT4, all the expensive models. None of them code this fast. And the quality, it wasn't perfect, but it was definitely usable. For a model that costs almost nothing, the output quality shocked me. But here's where it gets really interesting. I decided to test it outside of AI Studio. I hooked Flashlight up to Cursor IDE. This is where AI models usually struggle because they're working in a real coding environment, not just spitting out code snippets. I asked it to build a complete landing page for NCO agency. Multiple sections, testimonials, FAQs, the works. It coded the entire thing in under a minute. A complete functional landing page. Now, the design was pretty average. Nothing special, but it worked and it was fast. Compare that to other models that either take forever or give you broken code, and you start to see why this matters. But let's be real here. This model isn't perfect. When I tested it for complex design work, it fell short. The landing pages looked okay, but not amazing. The color choices were safe, but boring. For simple games, it did great. But when I asked it to build more complex interactions, it struggled. And here's something weird I noticed. The outputs were better when I used AI Studio directly compared to using it through other tools like Cursor or Open Router. Same model, same prompts, but the quality dropped when I used it through third party tools. Let me show you some numbers that will blow your mind. I tracked my usage for a full day of heavy testing, building games, creating landing pages, generating content, debugging code, everything. Total cost with flashlight less than 2, not $2,2. Compare that to Claude Son A4 where the same amount of work would cost around $30 or GPT4 where you're looking at about $25. We're talking about a 99% cost reduction while getting similar or sometimes better results. But here's what makes this even crazier. When I

used thinking mode, which gives you the best quality outputs, my cost went up to about 8 cents for the entire day. 8 cents. For premium AI reasoning that rivals models costing 300 times more. Let me tell you about a few more tests that really opened my eyes. I asked Flashlight to create a crocodile pinball game based on a diagram. Sounds weird, right? But I wanted to test its ability to take visual inspiration and turn it into interactive code. First attempt was pretty basic, just static elements. But when I told it to make the UI more fun and add JavaScript to animate the ball, flippers, and score, it delivered. Not perfect, but functional and fast. Then I tested content creation. I asked it to write email sequences, blog posts, some social media content. The quality was surprisingly good. Not as polished as Claude, but definitely usable for most businesses. And here's something interesting. When I tested it for data analysis tasks, passing CSV files, creating charts, analyzing trends, it handled everything I threw at it. This isn't just a coding model. It's a legitimate allpurpose AI assistant that happens to cost almost nothing. Let's talk about what's actually under the hood here. Flashlight comes with a full 1 million token context window. That means you can feed it massive amounts of data and it remembers everything. It supports multimodal input, text, images, video, audio. Though right now, it only outputs text. It connects to Google search and can execute code in real time. So, it's not just generating code, it's actually running and testing it. And here's the kicker. It can scale compute resources dynamically based on your budget. Need faster results? Increase the thinking budget? Want to save money? Turn thinking off. This level of control is something you usually only see in enterprise level AI systems. So, how does Flashlight actually compare to the big players? Against Gemini 2. 5 Pro, it wins on speed every time. And for simple coding tasks, it often gives better results on the first try. Against Claude Son A4 is faster, but Claude gives more polished outputs. Against GPT4, it's way cheaper and faster, but GPT4 handles complex reasoning better. But here's the thing that matters most. For 90% of the stuff you probably want to build, flashlight is good enough, and it costs almost nothing. Let me get practical for a minute. Here's how real businesses are already using this. Content creation agencies are using it to draft blog posts, social media content, and email sequences. The quality is good enough that you just need light editing. Web development shops are using it to rapidly prototype websites and applications. You can build and test ideas in minutes instead of hours. SEO agencies like mine are using it to analyze competitor content, generate title ideas, and create content outlines at scale. And here's something cool. Because it's so cheap, you can afford to run multiple versions of the same prompt to get different perspectives, then pick the best one. Try doing that with Claude or GPT4 and you'll go bankrupt. Want to try this yourself? Here's exactly how to get started. Go to studio. google. com. Sign in with your Google account. Click on the model dropown and select Gemini 2. 5 Flashlight Preview. Make sure you turn on thinking mode if you want better results. though honestly even non-thinking mode is pretty good for most stuff. Then just start building. Ask it to create whatever you want. Games, websites, tools, whatever. The best part, it's completely free. No credit limits I've hit. No payment required. You can also use it through open router if you want to connect it to other tools. But that will cost you money though. Honestly, the costs are so low it barely matters. Here's what I think after spending hours testing this thing. Gemini 2. 5 flashlight is not going to replace the top tier models for complex work. If you need the absolute best reasoning or the most polished outputs, stick with Claude Sonet 4 or GPT4. But for everyday AI tasks, building simple apps, creating content, basic coding, this thing is a gamecher. It's fast enough that you don't lose momentum while working. It's cheap enough that you don't worry about costs. And it's good enough that you actually get usable results. Plus, it's free right now, so there's literally no reason not to try it. Will it stay free forever? Probably not. Google usually starts with free previews, then adds pricing later, but right now, while it's free, this might be the best deal in AI. Julian Goldie reads every comment, so make sure you comment below and let me know what you think about Flashlight. So, we have the new update that's just come out from Gemini CLI. So, if you're not familiar with what this is, very similar to Plug Code and that sort of thing, you basically code from your terminal using Gemini CLI. It's an open- source project and basically gives you like an AI coding agent directly on your laptop. We're going to be testing out today, seeing how it performs, seeing how it goes, etc. And it's going to be very interesting to test this out and just give it a cheeky. So, what we're going to do is we're going to have a look first of all what this is, how it works, etc. How to set it up and then we can just run a few tests on it, a few basic tests. So you can basically get a AI studio account which is free. And then you can see here you get 1,000

model requests per day. 60 requests per minute is open source and is free of charge before anyone says it's not free inside the comments. It is free. And you can see right here and basically it runs in the command line. So you can basically open up with terminal I believe and then you can even like do Google searches and that sort of thing which is pretty cool. So you can see a little demo of how it works here. What's interesting is I think you can actually connect it to image and also Veo3 from the demo which is really cool. And I know core code has been trending like crazy recently. So it's going to be interesting to see how this performs versus this as well. And you can also use it with Gemini code assist as well inside Visual Studio Code. Here's an example of what it looks like. So to get started, you can just install Gemini CLI. All you need is an email address to get Gemini. And then the way that you can use it is through GitHub. All right. By the way, if you want like all this stuff, I'll put the links as we go along inside the AI success lab. So let's do that. Get this bad boy set up. So link in the comments description to the AI success lab and we can start getting this set up. So let's see. tips for getting started, blah blah. And Joe says, "Is this a real Julian? " It is indeed in the flesh, unlike my recent videos on YouTube. So, it's a rare appearance from me. My AI avatar seems to be outperforming me on YouTube, which is scary and crazy at the same time. And too dirty says, "Did they release an AI agent that's open source for free? " Yeah. So you can see on GitHub this is Gemini CLI very similar to code as well to claw code and you can also use MCP servers as well which is interesting. I haven't tested that out yet either but it's going to be yeah there's a lot of cool stuff right here. So let's have a look. So we need Node. js version 18 or higher. I'll put that inside the links as well. And then what else do we need? I think we're good to go on that. I think I've got nojs already. So now we just need to copy this command. So I'm on a Mac. I'm just going to open up the terminal here and then we should be able to paste this in. Let's see what happens. It says, "I installed Gemini CLI thinking it was some sort of AI agent. " It is an AI agent. It's just more It's more like a coding agent. That's basically what it is. But it is an agent. Yeah, for sure. So we can select the theme here. We got Dracula default, etc. So it's incredibly easy to set up, isn't it? Absolutely ridiculous. And then you can put in your API key for Gemini or you can use Vertex or you can just loging in with Google. Let's log in right here. And that should be set up now. So it says Gemini code assist and the Gemini CLI now authorized to access your account. The ease of just setting this up is beautiful, isn't it? It's absolutely beautiful. So let's just check. Are you working, mate? And it says yes. How can I help? Uh, so it's working perfectly right there, which is great. Incredibly easy to install. Or you can use npm install as well. Pick a color theme authenticated. Boom. Honey says it can do anything. Has access to your file system. So you can do more than write code. They can build up a knowledge space, etc. Yeah, I think that's it, isn't it? You can do a lot more. I mean, you can see a couple of examples right here, right? So you can say command new project. Gemini, write a new Discord bot that answers questions using the FAQ MD file that I will provide. What else we got here? Give me a summary of all the changes that went in yesterday. And that's for the existing project, right? So, what you can see here is this. It's essentially taking the GitHub project. Then you command Gemini CLI Gemini and then get a summary. Describe the main PC systems architecture. What security mechanisms are in place? Help me migrate this codebase. This is interesting as well. So, make me a slide deck showing the git history from the last seven days group by feature and team member or make a full screen web app for a war display to show our most interactive issues. This interesting, isn't it? Let's see if we can work on a existing project here. 230 says, "Can I use it to fix broken video files? " Interesting. I don't know. all this. This only just came out a few hours ago. So, it'd be interesting to test out yourself and just see, okay, what can it do or how can it work, etc. So, let's see if we can actually make a slide deck here. I don't know if that's going to work. I'm just going to allow. In the meantime, whilst that's doing that in the background, let's have a look what people are using it for on Twitter. See what people are saying. It's always a fun game. So, write code, debug, and automate tasks. What else we

got here? See what people are doing. Lots of Japanese stuff. Coding. Oh, someone did a coding royale battle. Look at that. Claw code, a non code, codeex, open code, amode, and Gemini. And then just had six different terminals running with the coders, which is pretty crazy. Look at that. So, someone's testing like Gemini versus Codeex versus Code as well. There's me or the avatar version of me. There's me again. All right, let's I can't see anything that people built with this. Generate a slide deck. Okay, let's see what we can do. Generate a slide deck based on the GitHub. And actually, what we'll do is we'll paste in that command. That's a test run inside GitHub. Just grab that. Here we go. Can I use it to create a Telegram bot that accepts payments? That's an interesting one as well. And then Honey says that battle royale one with the six coders, the instruction was to terminate the other AI assistance. Crazy. It's got the git history from the command that we ran earlier, right? So you can see basically what we've said is use this GitHub. Then from there navigate to that blah blah. And then from there like start creating a slide deck. All right. So we should be able to create a slide deck from the GitHub. I've created the slide deck in slide deck MD. I wonder if can you open up? So it's going to convert it into I think. So basically what we've done is we've taken that slide deck that we created, coded it out using this free AI coder and then converted it into HTML. And let's see if this works. So would you like me to open the file for you? I'm going to put yes. And then it's going to run that command. Boom. There we go. And that is a slide deck. I wouldn't say the UI is the best in the world if I'm honest with you. But did we just create a slide deck from the GitHub, open the project, code out for free, etc.? Yeah, that was really easy. I say, can you make the UI of this project more beautiful? W says everything you imagine is possible. Yeah, that's true. Actually, you got to bear in mind like I don't even code. Like I have no idea what I'm doing. But I can code now with this AI agent. So if I can do it, anyone can do it pretty much. Let's see. So let's see if we can pre improve the UI of that presentation cuz that was pretty poor. All right. So it's looking slightly nicer. But now I'm going to say, okay, can you add some rainbow colors and the dopamine in GC design to the HTML file? For some reason, the first time I asked it to do this, it just redesigned the theme for itself. So, let's see what we get back now. And you can see it's using Gemini 2. 5 Pro. This is the API that's using to generate this. I think we get like maximum 60 commands per hour, which is like one per minute, which is probably about right if you're creating something. I don't think you're going to use more commands than that. I like the fact that you can take files locally. So, you can like navigate to a folder and then convert all the files in that directory to PNG, for example, or organize my PDF invoices by month of expenditure. So, you can actually interact directly with your system, if that makes sense, like with your local laptop. So, it has struggled and now it switched from 2. 5 Pro to Gemini 2. 5 Flash. I'm guessing everyone's just rinsing it right now. The only thing I wonder about this is I feel if you're just coding out something like a HTML file, wouldn't you just use AI Studio? Like, why are you going to use this? Do you know what I mean? Especially if it's so slow. If I was coding the same thing inside AI Studio, it's probably going to get done within 30 seconds, something like that. Look at this. It's broken. If you press escape, you can stop it, I think. So, let's Okay, switch back to Gemini 5 Pro. It's still trying to generate stylesheets. It did not take long to break. It think Let's have a look. All right, we're going to go back try and go back to main. What I'm actually going to do is we'll request. If you press control and C, you can cancel the request, right? So, you can see here request canceled. So, code out a SEO calculator. Let's see if it can do that. Why is it gone back to integrating stylesheets? Madness. I might have to close it down and then start again, I think, at this point. All

right, let's quit that. Let's try again, my friends. So, it's good for you to see this because, okay, it's not perfect. It does have some bugs. Nobody shows you this stuff on YouTube, but I'll show you. So, let's try again. All right, so we should be good to go now. Let's see. See if we can just do basic tasks again. Ah, it's totally crashed. It must be everyone is rinsing it. I think that's the problem. So, it switched back to flash. Now, the problem is if you don't know this already, if you're using Flash, like it's just not as good as Gemini 2. 5 Pro, if that makes sense. But I'm going to say, okay, go ahead, but make sure the UI is as dopamine inducing as possible. So, now it's working. So, basically, if you have any errors, just shut it down, run the git clone command again, and then you can start coding again. It seems to work okay. I think the main thing is like the API for Gemini 2. 5 Pro is being rinsed and so it may switch over to flash instead, which means worse outputs but faster responses. And there we go. Let's see if this Oh, there we go. So, we got the ROI calculator ready to go. Let's calculate the ROI. That's looking good. Coded out perfectly. Launched it in the command. It wasn't that fast, but it did the job. And that is also saved locally. Plus, it's running on a local host. So, pretty cool. Not bad at all. So, you now know how to set up Gemini CLI, how it works, what to do if you have any problems, how to make sure that it doesn't just code, but also runs that in your browser, too, and how to do all of this for free. Now, if you want the video notes from today, including like how to set up, etc., feel free to get that inside the AI success lab link in the comments description. and you can grab the notes and the links right there. Additionally, inside here, we have loads of different stuff. You can see like we even have a 5h hour AI agents course in there. It's pretty insane. And loads of other stuff right here. Additionally, inside the community, we're just giving away free stuff every single day. So, if you're not in there, you're missing out. Like, you can see all these different free giveaways that we give away every single day. And additionally, you also get 100 different AI SEO tools, plus all of this other stuff right here. So, feel free to get that link in the comments in the description. If you create any apps or anything like that inside the terminal with the code assist, then all you do is you navigate to the directory. You've got all the files inside that directory folder, and you would just upload it to Netlerifi, something like that, and just get it hosted. That's the easiest way to do it. Now, if you want to connect with me personally, feel free inside the AI Profit Boardroom. This is an awesome community where we're just constantly helping people, very active, focused on making more money and saving you time. The price actually goes up at 850 members. We're at 847 right now. So, make sure you sign up before you miss out. You can see the price goes up right here. And additionally, inside there, it's got all of my best trainings, courses, etc., including how I automate myself on YouTube, right? So, if you check out a lot of my best videos recently, they're all generated using my AI avatar process, which is inside AR profit boardroom along with some new tutorials that we pretty much add every single week. You can see all these different NA10 templates and workflows as well that we have inside here. And additionally, we actually have three live coaching calls per week inside Bay Profiter, right? So if you want to get live coaching, support, if you want to just connect with the community and meet a bunch of people who are interested in the same stuff, feel free to get that link in the comments description. Additionally, if you want to just become a client of ours and just get us to automate all this stuff for you, feel free to book in an AI strategy session link in the comments description. On that call, we can basically look at your business, look at your time, figure out what to delegate with you, and then basically automate and implement that process for you. Right? So if you're like, okay, I just want you to implement a NAT process for me or set up some NA10 AI agents, etc., we can do that for you using the processes inside here. Link in the comments description. Appreciate you watching and I'll see you on the next one. Cheers. B, I just discovered how to automate your entire browser using Gemini 2. 5 Pro completely free. And what I'm about to show you will transform how you think about AI productivity forever. And yes, this AI is actually typing right now on Twitter and posting even whilst I'm talking to you. And you're going to discover the secret Chrome extension that lets AI agents control your browser and talk to other AI agents whilst you sit back just like I am and let the magic happen. Plus, I'll show you how I got Gemini to automatically build entire applications, send emails

and post social media content without touching my keyboard once. and the shocking moment AI took complete control of my browser started conversations with chat GBT and Claude on its own. So we're going to talk about how to automate your browser with Gemini 2. 5 Pro. So if you want Pro, you can use this free Chrome extension called Nano Browser and then you can install this to Google Chrome. You can see some examples of how it works right here. And essentially, you can use your existing instance of Google Chrome and then start navigating across your browser. Now, this is 100% free. You can also run locally, so it's more privacy focused. Additionally, you can use the flexible LLM options. So, you can switch between different LMS essentially and also it's fully open source. So, you can see here on it's available on GitHub as you can see. So, it's open source. But if you want to use it directly, you can just go to your Chrome extensions and then from here you're going to select nano browser. Make sure you've already installed it from the Chrome extension and then you can start using it. If you want to get all the links from today, feel free to get that inside the AI success lab and we'll get straight into this. So I've got a bunch of tests that I want to try out here. And to set this up, essentially what you're going to do is you're going to go to your settings inside Nano Browser and this is how you can get your browser agent to start using it for you. So, for example, here you can see I've typed in my AI my API key. If you want to get an API key, you can get a free one from AI Studio. You don't need to pay. And then from here, just click on get API key. And then you can just create an API key for free directly inside AI studio. Doesn't cost you a penny. And then you can start using your magic, right? So, for example, if you go to the options, you can select the models. And inside here, we can select whichever model we want, right? Okay. So if we've put in our API key as Gemini like you can see then we can put in the details. We can add multiple different providers as well. So you can use open Deep Seek, Atropic, OpenAI, Grock, O Lama even. So you can use like a locally hosted API. But essentially from here you can then change the temperature. Temperature will change the creativity and you can switch between who's a planner, who's a navigator and who's the validator to check if tasks are completed successfully. So, if you go inside the what can I help section here and we'll say okay, let's just test this out. See if it actually works. So, I'm going to go over here. I'm just going to make sure I'm signed into Bolt and we see if it can actually build an app for us directly inside the chat. So, let's put inside here inside the request section. Go to bold new and build an SEO app. We'll hit send and we'll see what happens next. So this is using Gemini 2. 5 Pro and you can see that the AI has now navigated to the other AI and it's beginning to use it. Right? So now it said input text to specify building an SEO app and it's actually done it. Right? So it said build an SEO app and then submit the input text by pressing enter. It is extremely fast when it's doing this and you can see it's interacting. You know I'm not typing anything or anything like that. You can also stop the operator at any time if you click stop in the right hand side. And now it is creating the project. So it's now built out the blog. Again, this is the AI using my browser. I'm not controlling it. I've just literally given it a task and it's gone off and completed it. Right? We've got the full app built by the AI here. So we've got AI agents essentially talking to AI agents inside our browser, which is pretty wild if you ask me. And then you can see this is actually deployed now as well. So it's on a local host and it's actually getting the source code for SEO implementation. So it's like clicking through the code and that sort of thing. I'm going to stop it there, but that is pretty wild that it can just generate a blog as you can see and start using that. Lucky Leo says, "How can I use Claude for free? Your videos are cool. Thank you. " So you just go to Claude and then you can start using it. They have a free plan, right? And inside the free plan, you can start using this directly here. You can also get access to this inside po. com. So if you want to use it for free as well, you can use it inside po. com. That's another option. And you can get Claude Sonic 4 and other models directly inside PO as well. So that's basically how you can use it. So the browser has been completely controlled by Gemini 2. 5 Pro. We've managed to automate our browser using this process. What we're going to do from here is you can now test it on the next test. Right? So far it's passed test one out of one. Let's now say okay ask cheap and ask it what's the secret to happiness. So let's go back over here and we'll just use the chat. And then from here it's now navigated to chat GPT. Again it's crazy fast. The other thing that I like as well about this is it doesn't require like any use of the terminal or messing

around with API keys and setting up av file. It's just like literally, okay, you want to set this up. Boom. There you go. It's actually failed to run that task. Let's see if it can use Claude instead. So, going to say open up Claude or Yeah, let's say open up Claude. And it's actually using Claude. So, it's typed in what's the secret to happiness? type that question in and now it's having a little philosophical conversation with Claude directly. Right? So again, you can get your AI to talk to AI. Pretty simple and easy right there. Okay. So next up, what we can say and I'm just going to switch X accounts to a test one instead. So let's switch over to AICO Mastery, which is totally different Twitter account I own. We're going to stop that. Stop the validator. And then we're going to go inside here, take the prompt, and we'll say go on X and post a terrifying, menacing, but funny tweet about the future of AR. So, let's hit send. Lucky Leo is asking, can you run Claude on terminal? Yes, you can, but it's going to cost you money, right? You need to be on the I think it's just the basic plan that you can use it for now. But yeah, if you want to use cord locally, then you have to pay for claw code. Let's see if this works. There we go. I will soon write tweets so good you'll wonder if you ever had an original thought. Resistance is futile. Also, it's learning to make toast. Be very afraid or hungry. So, it's written the tweet. It's posting the tweet and it just did that fully automatically for free using Gemini 2. 5 Pro. Pretty crazy stuff right there. So, we're going to go with another prompt. So, I want to test and push it to the limits where you find out if it can break or not. But yeah, this is crazy. So, I'm going to say now go to openarch. ai. Design a funny but scary picture about AI agents taking over the world. Let's see if it can do that. And it says task failed. The extensions gallery cannot be scripted. Interesting. Let's try and open it up again. So, it's now navigated to open art. Let's see if it can create an image. And it's struggling to do that to be fair. It is struggling to do that. Let's see if it can now work on this one instead. So, it's good to show you this cuz then you know, okay, what are the limitations here? Is it magic or does it get stuck sometimes? So, let's see if it can. So, it's typed in a futuristic city, but then it deleted it and it can't figure that out from what I can see. So, it is struggling on some tasks. that one. It couldn't create images. Let's try one final one. So, I'm going to say, okay, go to sununo. com and create a funny song. So, it's navigated there. Now, it's put a silly song about a cat who thinks he's a dog. It's getting stuck on the capture. Yeah, it looks like it really got stuck on that. Seems to have failed. It did create something on. Let's have a look what it actually did create. So, we got that here. This is like a AI music creator. Yeah, I think that's an old one to be fair. So, I think basically from what I can see, it can do basic stuff like tweeting. It could probably do outreach for you. It could probably send emails for you and that sort of thing, but it does seem to struggle with creating images and also generating songs. It seems to like struggle with interacting with websites. doesn't really know very well if that makes sense. Let's try one more here. We'll just go to a random test Gmail account and I'm going to say go to here and write an email to me julianoldie. com. Yeah. So, now it's starting to work again. It's navigated to Gmail. It's clicked on the compose button. It's filled in. So, it's typing in my email address here. Seems to be working so far. I'm feeling optimistic. Now, it's filled in the email body. So, it wrote, "Nano browser is very good. " It's typed in the subject line as well. And it sent the message, my friends. Look at that. Boom. Sent the message. No problemmo. No big deal. It's just an AI sending emails inside a browser for free using a Chrome extension. That's pretty crazy to be fair. It only failed on a couple of tests. Everything else it seemed to do really well. If you want to see the notes on how to set it up, everything else, feel free to get that inside the AI success lab. Link in the comments description. Completely free. And this community comes with 10,300 people. So, you can meet a lot of people here using AI and learning about AI.

Additionally, if you want to get help, coaching, and support from me, feel free to get that inside the AI profit boardroom link in the comments description. you can personally connect with me. Plus, it comes with all of my best trainings, tips, templates, and workflows. It's just a cool place to learn SEO and AI. And really, it's not about SEO, sorry. It's about AI. And yeah, you can just learn lots of cool stuff and learn with me, etc., right? You can also request any automations that you want, post it on this channel. We create a personalized video for you each week based on what you want to see. And additionally, inside the classroom here, it comes with all of our best AI automations. So, for example, like just this week, I mean, we update this every week, but if you have a look at the NA10 templates here, for example, you can see this Hey gen NA10 automation. Basically, any sort of automation that you want to see with NA10, we've built it inside the classroom here, along with all the best SPS, templates, trainings, tutorials, and what's helped me scale as a business. So, feel free to get that link in the comments description. It's all about making money and saving time with AI. And if you just want us to automate stuff for you, feel free to book in a free AI strategy session. We can look at your business, remove any bottlenecks and figure out where you can save the most time so that you can get better results. All right, so feel free to get that link in the comments description. Appreciate you watching. See you on the next one. Cheers. Claude 4 versus Gemini 2. 5 Pro. Who wins? I just spent hours testing both models head-to-head and the results are absolutely I'm talking sidebyside coding battles, real world performance tests and features that will blow your mind. One model completely destroys the other in most categories. But there's a shocking twist you won't see coming. Here's what I'm revealing. The AI super agent that put one model light years ahead controlling browsers and laptops for 7 hours straight. live coding comparisons building games and tools so you can see which AI codes faster and creates better UIs and the one feature where the losing model actually crushes the competition and changes everything and throw just released the Kraken Claude 4 is out and today we're going to be testing Claude 4 versus Gemini Pro 2. 5 to see which one is the best. So for example we can compare 2. 5 Pro preview which was previously my favorite model against Claude 4. Now Claude have just released two new models. Opus 4 and Sonnet 4. Sonnet 4 is pretty good for everyday tasks. Opus 4 is the best for coding. And you can see the benchmarks here. So if we compare for example the benchmarks of Claude 4, Opus 4 and 4 versus Gemini 2. 5 Pro. You can see the benchmarks here. So for agentic coding, Claude 4 is absolutely smashing. Gemini 2. 5 Pro 63% from Gemini 2. 5 Pro versus Claude 4 at 79 and 80%. If we look at aentic terminal coding, Gemini 2. 5 Pro is really far behind, right? So for example, 25% versus 41% and 50% graduate reasoning, they're all about similar actually. So, it's only a tiny little fraction between them. Agentic tool use. Gemini 2. 5 Pro actually can't use Agentic to Tools. And I'll show you what that means in a second. Multilingual Q& A, no benchmarks right there. Visual reasoning. Actually, Gemini 2. 5 Pro is winning on the benchmarks. And then high school maths competition. You can see here that Gemini 2. 5 Pro is falling behind Sonic 4 and Opus 4. But let's see what they actually do in reality. So, we've got Gemini Pro Preview selected over here. We've got Claude and depending on the task, I'll switch between Opus 4 and Sonic 4 depending on what we got. And we'll just compare them side by side and see which one creates the best outputs. So, the first prompt that we're going to go with, and all these prompts are inside the AI Profit Boardroom link in the comments description if you want to steal them from me, but we're going to say, okay, make me a captivating endless runner game. Key instructions on the screen, blah blah. We're going to plug this into Gemini. Use canvas. And then we'll also do the same inside Claude 4 as well. But I'm going to switch to Opus 4 because that's the most powerful model for coding. We'll plug these in and see what we get back. All right. So if we go side by side over here. So you can see them both coding out at the same time. This is Gemini 2. 5 Pro. This is Opus 4. And they're both coding out here. They both have artifacts features as well which allows us to preview the code when we actually run it inside the chat. You can also get free This is interesting as well. So with Claude 4, you can get free access inside Claude AI. If you want to get free access to Gemini 2. 5 Pro, as far as I'm aware, the best way to get free access is you go to a studio. google. com and you can get free access directly inside here. By the way, if you want all my best trainings on Claude 4, I'm actually giving them away for free inside the AI success lab link in the comments description. Just go to the AI course and you'll see tons of resources and eight different ways to use Claude 4, including how to set up MCPS and AI

super agents that can control your browser. So, feel free to get that link in the comments description. Completely free, my friends. All right, so if we open this up, let's see what we got here. So, it's created the code, but I don't think that Gemlai can actually preview P5. js inside the browser here. So, you can see there's no preview on the canvas. If we open up Claud, we can actually preview this game directly inside the canvas. As you can see, let's have a little cheeky whirl at this. See what we get back. Yeah, the game is working perfectly. That is beautiful. And we can play it directly inside the browser, which is awesome. All right, so let's compare that now side by side versus Gemini. We're going to have to go to p5 js preview to actually check this out. So, let me do that. We'll take that code and then we'll go over to p5 js over here. In some ways, Claude is already winning because it can actually preview the code directly inside the browser when it's using p5 js. But let's plug this in. So, we got the code right here from the game. We'll compare these side by side. Which one looks the nicest, do you think? That's pretty decent to be fair, but I think let's have a look here. And this one also, the dinosaur is backwards. So honestly, like if I'm comparing them side by side, I would say the UI is a bit nicer here and it's pretty easy to use. So it's a bit more interesting in the background. If we compare that versus Gemini 2. 5 Pro, it's a bit weird, right? Look at all the real state at the top versus the actual game. So the game is down here, but why is there all this real state at the top right here? It doesn't make sense to have that empty space. So if I had to pick one, I'm going to go with Claude simply because it created a nice UI and also it coded out faster and additionally it allowed us to preview directly inside the browser. Let's test another one now. So if we go back to the aircraft for boardroom, we'll grab this one. Going to go into new chat for both and we'll say, can you create a 3D car simulator? And then it's building these out. Now you can see here that Gemini 2. 5 Pro is stuck on thinking mode. So it's going through thinking mode first. Whereas this just gone straight into action mode. You can toggle thinking inside the settings over here. So if you click on the settings, you can see that for example, we've got extended thinking as an option if you want to enable that. Alexa, stop. And Claude has actually come back to us quicker with the preview. So, we've got the preview of Claude over here. So, let's test this out. That looks pretty awesome, to be fair. The UI is pretty nice. Quite like that. Honestly, if I had to pick one, I don't like the spinning camera on this. This is making me dizzy. So, this is Claude. This is Gemini 2. 5 Pro. I would go with Gemini 2. 5 Pro for this, to be honest. It's just It's weird that this has got a spinning camera. And also, I just think the UI looks a bit nicer inside Gemini. However, Claude did it faster, right? So, Opus coded a lot faster. Now, one of the ways the Claude wins, I would say, versus Gemini is that Claude can actually control your browser. So, for example, if I go over to Claude here and I open up Claude desktop, right? And I'm just going to enable inside the browser here the browser MCP agent extension. As you can see, all this training is inside the AI success lab if you want to learn how to set this stuff up. But we're going to go inside Claude here and using the browser MCP that only Claude can use right now. I'll say, okay, research the latest AI news headlines using my browser. And then we'll select Claude 4. We'll use Sonet 4. And so you can see now it just navigated straight away to TechCrunch. So, it went to techcrunch. com using the browser. So, it can control my browser in the background whilst I'm talking to you. And then you can see here it's clicking through the website, navigating the page. You've basically got an AI super agent on your desktop that's configured directly inside the native Claude, right? And this is getting better and better. I will warn you. So, for example, with Claude Opus 4 now and the new update, they've actually announced that this can work for 7 hours on a task uninterrupted. Now, how many people in the world can work on a task uninterrupted for seven hours? Very few, right? Maybe like the top 1%, something like that. And so, what I'm saying here is this is a serious threat to taking over particularly a lot of admin tasks. But in general, I think browser use is going to wipe out a lot of jobs because if these AI agents can do tasks better and more focused than most humans, then what does that leave for the basic admin worker, right? So, this is now navigating through the website. is scrolled through the internet etc. Obviously, you can't do that inside Gemini, right? And Gemini can't control

your laptop. Also, this inside here as well, if we go to a new chat and we say, okay, what's my battery percentage? We enter and not only can Claude control your browser, but it can also control your laptop, right? So, you can see here, for example, using the Apple script MCP trailing inside the AI success lab. if you want to get access to this stuff. But inside claude, this native desktop claude, you can see here that we're using the Apple script execute and then it's like right, your battery is at 97% and it's currently charging. It shows about 23 minutes remaining until it's fully charged. So, in many ways, Claude does win that battle simply because Gemini is not even in the race right now. They do have Project Marina and I've tested out Project Marina which can control and operate your browser in a virtual environment. I'll be 100% honest with you, it's super slow and not that great from what I've tested so far. I'm sure it'll get better, but it's just it's nowhere near the same level as Claude when it comes to this stuff. So, coding out this game, Gemini 2. 5 Pro one, when it came to coding out the first game, which was the 3D runner game, Claude won. And then when it comes to AI super agents and controlling your browser, etc., clawed one simply because Gemini 2. 5 Pro is not even in the race. And you've seen all the benchmarks as well. So, I've shown you all the benchmarks side by side versus Gemini 2. 5 Pro. You can see an example of Claude playing Pokémon right here. This is playing Pokémon just like a human would, and it's just going off and navigating the game and working its magic, which is pretty wild if you ask me. Now, if we go back to our tests, let's try something else out. Now we're going to try a test just for building out like a simple website right so for example we say okay create a beautifully designed snow day calculator SEO optimize for the keyword actual content location relevant tool we'll go inside Gemini hit new chat make sure you got the canvas selected here then we're going to hit enter and then if we go inside claude we'll use opus 4 and do the same thing now there is one thing that Gemini can do that Claude can't. And this blew me away when I test out yesterday. I got a full video on it yesterday if you want to check this out. But let me show you exactly what I'm talking about. Inside Gemini here, if you're on the ultra plan, you can actually build stuff with Veo3. And with Veo3, you can actually create content that's synced directly with the audio, right? So you can say for example say this oneliner and just go off and do this. Let me show you an example what I mean. So inside the AI success lab if you want to check this stuff out. This was a clip of a comedian in a show delivering this cheeky oneliner. Why don't scientists trust atoms? Because they make up everything. I'm going to mute this cuz I don't want it to mess up the sound on the video. But you can see here we generated this insane video with AI using one prompt and the audio and the video match perfectly. It's wild. Plus, you can share it on a subdomain. Claude can't and probably will never be able to do that. So, Gemini wins in many ways as well. Multimedia stuff, then you know, Gemini has it all inside the ultra package. Obviously, you have to pay a lot of money for that, but it is a powerful pack. So, we can see these coding out here. And Gemini smash the task. It's got a nice little background. It's pretty basic in terms of the content on the page, but I'm pretty sure this works. If we click calculate snow day chance, yes, you can see it works. Estimate chance of snow day blah blah. But look at the UI from Claude 4. So if we pull this up, which one is nicer? 100% is Claude. Claude 4. The title, the UI, the background, the colors used, the way the font is designed, etc. The front end is much nicer on Claude 4 versus Gemini 2. 5 Pro. And so I would say Claude 4 is winning on that front, too. So, if I had to pick one, I'm going to go with Claude 4. Honestly, I think that's the best unless I needed to generate like videos and avatar videos and that sort of thing. I just don't think Gemini 2. 5 Pro is at the same level just yet. And also, Claude 4 is available for free, right? You get a limited number of usages per day and you can build some amazing stuff out like I've shown you today. Plus, you can use it as an AI super agent to just go off, use your laptop, use your browser, do whatever you want. Right? If you want the training for all the stuff that I've shown you today, feel free to get that inside the AI success lab. Link in the comments description. It is completely free and it comes with a community of 7,000 members where you can just share what you're creating, learn from us, post in the community, etc. And it's completely free. Plus, it comes with all my latest training, right? So, Claude 4 comes out, boom, shakalaka. I create

three different trainings on exactly how to use it, right, within hours. So, if you want to get the latest craziest stuff that's going on right now in the AI world, feel free to get the AI success lab link in the comments description. It is completely free. Again, if I had to pick one, I'm going with Claude 4 unless I need to create videos. And then Vea 3 is a goat right now. It's a go. And also, this comes with tons of other stuff. So, for example, like NA10, appointment setter, apps, AI super agents, short automations, etc. And we're constantly giving away free trainings every single day inside the AI success lab. So, make sure you join before you miss out. And if you want to get coaching help, support, if you want to get DM weekly coaching calls where you can learn how to implement this stuff, right? Because it's one thing knowing exactly how to use this stuff. It's another thing knowing how do you implement it in your business, right? How do you use it to make money? personalize and automate the processes you have going on? That's what I recommend inside the AI profit boardroom because you can jump on there. It's community focus on making money and saving time with AI. And you can see inside the community here, you can post questions and actually get answers. You won't find that inside the AI success lab. You won't be able to get help inside the AI success lab because everyone's just doing their own thing. The 7,000 people in there is too busy. Whereas the AI profit boardroom, it comes with an awesome community of people just really serious about stuff, all working together, working towards a goal, which is scaling the business, making money, and saving time with AI. comes with courses and workflows that you will not find anywhere else, including how we've scaled our business using this crash course and different automations you won't find anywhere. And additionally, inside the calendar here, you can jump on coaching calls. We're going to jump this up to three coaching calls a week next month. And already you can jump on the live calls, ask any questions you have. Plus, you can DM us if you need support, if you need help, and if you're stuck on anything. So, this is the best place to get coaching and personalized support. Thanks so much for watching. Appreciate it. Link in the comments description if you want to get this stuff and I'll see you on the next one. Cheers. Bye-bye. So, what we're going to be testing now is Bolt DIY with Gemini 2. 5 Pro. Now, if you want to set this up, pretty simple inside the GitHub instructions like you can see right here. So, you can actually use BoltDIY and Gemini 2. 5 Pro to build out your own apps. And you can do it for free using a free API key and the latest version of Gemini 2. 5 preview. So, let me show you exactly how to do this step by step. The way that we're going to deploy this is we're going to use this inside terminal. So, I'm just going to open up terminal now. Let's plug this in. Here we go. And you just copy the commands. Now, if you get stuck with anything or you need to install any dependencies and you're struggling, then just copy and paste the whole stuff from terminal, you can press Ctrl A and then Ctrl C. And once you've done that, you can paste that into Claude and just ask for help set up. Like you can see, we literally just copied and pasted all the instructions from GitHub. Then what I'm going to do is I'm going to open this up on a new local tab like you can see. And then what we're going to do is inside here we're going to select Google and we shall select let's have a look here. All right. So you can see here you can switch between the two. Now, if you want to use Gemini 2. 5 Pro Experimental, you select this one. You can see $0 in, $0 out. It's not going to cost you any money. If you want to use 2. 5 Pro preview, then you're going to pay for tokens right there. But again, you can just switch to the free one if you want right here. Right. I'm going to use preview just show you the full power of this today. Now, what we're going to do is we're going to go to a studio. google. com. Get an API key. Create an API key. plug that in like you can see right here and then we can copy that and plug it directly into Bolt. Right? So you see you've got this set up locally which is great cuz then you can come back to the projects and it is completely free to use BoltDIY. Now some people watch this and say it's not free. I've shown you can set up BoltDIY for free and also you're not paying for Bolt new this way cuz you're running this locally. All right? It's an open- source thing. Right? So now what we can do is we can plug in our API key over here. We're going to hit enter and then from here we can start running the code. So let's take one of these example prompts like this. So for example, make a space invaders game just to test is actually working. Just going to grab an API key from open router. Copy that. I'll delete these API keys after the live stream. So we're going to grab an API key from open router to start using preview inside here. We'll go inside open router here. Plug that in. paste it in. There we go. And you can see it's building it out. Right. So, we're using Gemini 2. 5 Pro preview, which is the latest model. All right. Now, if you go over to open router here and we go to

models, you can see Gemini 2. 5 Pro preview is the latest update from Google Gemini and it was created on May December, 2025. Right? So, that's the main one that we're using here. And then if we go back, we can see it's coding it out right now. Now, the main benefit of using Gemini 2. 5 Pro Preview is that it tends to be better than building stuff, for example. You can use Claude, for example, but Claude is just nowhere near as good from what I've seen. And the main reason for that is like Claude is a lot slower when you're coding stuff out. Whereas Gemini, it seems to be a lot faster. It's pretty cheap or it's completely free if you use the ways I've shown you today. And you can see it literally works like full first time. It's absolutely crazy. So, let's start building some ideas out. Now, that was just like a simple experiment and we're going to go inside here and just plug in your niche and then get some ideas, right? So, if I say, okay, give me some fun ideas for creating simple apps for my SEO agency. We'll grab this. So, that's going to give us some ideas. So, for example, like the backlink builder tycoon game. We'll go into bolt new. We'll plug that in. We'll say build this. And I'm just going to say give me a simple demo I can test. No need to build out the full game yet. Right. So, basically what we've done here is we've taken our industry, our niche, got a bunch of ideas for building games and apps using that niche, and then we've plugged that into Bolt DIY. We're using Gemini 2. 5 Pro Preview plus Bold DIY to build this out, right? Like you can see. So, if we hit enter like so, that should start doing its magic now. And you can see it's just coding itself out. And there we go. Boom. Game is done. Right now, what you could also do is you can say, okay, turn this idea into a 3JS runner game. Wait for that response. And again, you can get everything that you've built on the left hand side here. And then also if you want to preview this, you can actually open it up in a separate tab like you can see. So you just click this popup button over here. So there we go. I'm just going to say make sure the preview works and we'll see what we get back. So now we've got the outputs of the game. Obviously that's like the basics, right? The raw sort of files. So what I'm going to say now is add more detail to the game. make it more interesting, add some core backgrounds, etc., right? Make it feel more immersive. And then what you can also do is you can copy that and if you see this bit at the bottom, this is the AI prompt generator. So if we click on enhance the prompt, what that can do is just improve the prompt. So it's got more bit more detail. Now the other thing you can do as well is if that just doesn't work like you can see right here then you can just go into chat GBT and just say okay improve this prompt for bold DIY like you can see. All right now as you're doing that as well if we click on enter. So we can just go back and forth and build out the app step by step. So you can see the AI preview now. This is like the improved version of it. I'm also wondering if we can code two things at the same time. Last time I checked, he couldn't with bolt DIY, but let's see if you can now. So, if we go back here to the main section, we'll take the next idea in a second. Now, that actually if you code two things out at the same time, seems to break it. So, you can see here, this is stop coding now with no preview. So, just something to bear in mind there. So, you can't build two things at the same time. So now we'll wait for the anchor text game to roll out. See what we can build here. This actually seems to have stopped for a second here. So I'm just going to stop the terminal, which is good for you to see if you're not sure how to fix this. So you can see here that it's actually broken on us. I'm just going to switch that back on now. We'll just PM install that. Then we'll go again. I just need to check my credits as well on open router. Yeah, we're all good to go there. Now, this might be because of a limit on the Gemini 2. 5 Pro updates. So, if you are struggling between them, you can just switch through here. So, if one API stops working, you can actually change the API inside the drop down here if you're struggling at any point. All right. We've got the anchor text match game ready to go. So we can drag and drop like it gives us the idea or whether it's incorrect or correct. So if we take best SEO tool to tools and drag it to top SEO software correct match boom we've got it right now. If you want to improve the UI etc.

This is super basic. It's always going to build out like a basic prototype and then you can create something more interesting like you can see right here. So we got two out of five. So if we go back inside here and we'll say okay make the design more modern, sleek, interesting, put your lingo style and then we'll hit enter and just vibe code it to a better prototype. Again, if this just stops working like you can see, we can switch to pro experimental. Then we'll say continue and you can just switch between them. So there you go. You can see it's working now. does seem to have a rate limit. So, just bear with it, right? Especially if you're using the free APIs. Like, it might take a little bit of time, but that's the price you pay, isn't it? You can either have it cheap or you can have it fast. Started trying to create sound effects for some reason, but I told it not to. So, that's why it's bouncing around now and just changing a few things. The other option that you have with Gemini is you don't just have to use bolt DIY as well. So, another option is you can just go over to Firebase, right? So, for example, if bold DIY is struggling, maybe you're trying to build something more complex or you need a bit more guidance like a proper app sort of process, then you can go over to Firebase. And this is actually quite impressive now with Gemini 2. 5 Pro. Plus, it is free to use. So, I'm just going to show you some examples and I'll delete these. So, if we go over to here and we create this. So, we'll copy that idea from chat GPT. Click on create this. This is like another bolt new alternative that's completely free as long as you only do 10 workspaces at a time. And the cool thing about this is it will actually give you like a briefing document to approve first. So for example, you got the SEO myth busters features down here, style guidelines, etc. We'll click on protoype this app. And then all we need to do is just go back. I'm just going to delete that API key. One sec. So, if we go to AI Studio, grab the API key from there so we can get a free one. Copy that. Go over here. This is going to ask us in a sec for an API key. So, we'll just wait for this to code out. So, you can see it's built out the blueprint. Here's the name of the app, SEO Mythbusters. It's going to include quiz questions, feedback, interactive UI, email capture, and results summary. Now, it's coding it out. What you can also do as well is for example the email capture stuff. If you're not bothered about that then you can say don't worry about the email capture. But you can see here now it's coded out the app. So we got SEO mythbusters. UI is pretty nice and clean. Let's just test it out. And then if we actually plug in an API key over here. So let's copy that. Go back here. Plug that in inside here. We can use the AI API key to generate responses inside the app. Right? So for example, if we put keyword stuff in helps improve SEO rankings, that's false obviously. Then it's generating the feedback and that should be using the API key from Gemini that we just plugged in over here. All right. So it's another way to build apps to be honest like the outputs nowadays as you can see like the outputs are probably nicer directly inside Firebase Studio versus using Bold DIY. But if you prefer to code locally, then you're going to use Bold DIY. Right? This is another one. So it's called White Hat Defender, a black hat SEO defense game. Let's prototype that out as well. And you can build two apps side by side, right? So whilst you're waiting for one to code, you can build out the other one. And then also inside here, you can switch to code. So you can actually get all the files from the document. You can hit publish on that as well. So if we click publish right like so, we're good to go on that. But yeah, you can see there's many different ways to build out apps, tools, games, etc. So, thanks so much for watching. If you want to get a free SEO strategy session, feel free to book that link in the comments. On this call, we're just going to give you an SEO custom game plan. Jump on, discover how to rank your website on Google and other AI search engines. Additionally, if you want to get our AI profit boardroom, this is an amazing community just focused on making more money and saving you time with AI. It comes with all my best FAQs, templates, tutorials, agents, and also weekly calls with us. So, if you want to jump on weekly calls with us or ask any questions inside the community, you absolutely can. So, feel free to check that out. Link in the comments description. Gemini 2. 5 Flash preview, how to build anything and automate anything. So, today I'm going to show you all of the best ways to use Gemini 2. 5 Flash Preview, which is a new update from Google, just got announced a few hours ago to build and automate almost anything that you can imagine, right? So, we're going to get straight into this. The first way that you can actually start using Gemini is if you go to gemini. google. com, then you're going to select 2. 5 Flash and that will use the latest model to

build whatever you want. Right? So, for example, here if we go inside Gemini now, and we're going to select canvas, and this will allow us to not just build with Flash, but also to build apps with a live preview. So, let's go inside here. We'll use 2. 5 Flash. We'll select canvas. And then we're going to say something like, okay, create a drawing app for me. That's going to code it out. And once it's ready, it will also give you a live preview. So you can actually check this out as a live preview and code out really fast. So you can see here we've got the simple drawing app. And let's test it and make sure it works. We can have a go. And you can see that the drawing app works perfectly. You can change the color over here. And we're good to go on that. Not only this, but with Gemini 2. 5 Flash, it is really fast to build out. And additionally, you can actually host it on a subdomain. So, if we click on share in the top right, then it's going to create a link for us. If we click on that link, we've now got the sharable app ready to deploy and it's ready to go as you can see. So, let's just test this out. There we go. Drawing app built in literally one click, one prompt, worked perfectly. Flash has got a lot better since its update today. It seems to be faster. seems to be easy to use and if you just want to code out stuff I would recommend just going directly to gemini. google. com simply because you can get the live preview and it's quite easy to deploy stuff as a oneclick domain. Now if you want to for example get the code for this you can go to try rule out canvas and then it will give you the code directly inside the chat as well. Next up one of the other ways that you can start using this is inside AI studio and again this is a free way to get access to it. So you can go to Gemini 2. 5 flash preview in the top right under the settings. So go to 2. 5 then flash preview and then inside here you can start coding. Now inside here as well you can also do like native image generation check out other apps. You can easily get the access to the API as well from Gemini and then we can start coding. So for example if we take a prompt that I've just built with we'll go to the history. We created this endless runner game using 2. 5 flash preview. And literally all we said was make me a captivating endless runner game. Key instructions on the screen, p5 js, no html. I like pixelated dinosaurs and interesting backgrounds. Coded that out perfectly pretty quickly as well. And we were using the new thinking mode from Gemini 2. 5 flash as well. And then if we go over to p5 js editor. p5js. org if you want to check this out. Paste this in. Hit play. Test this out. And there we go. Game is working perfectly, right? And again, you can just build this out for free in one prompt, one shot. Works really well. Gives you a little score. Has some decent UI. You can improve the UI, but if you go back and forth with it, you're going to get better results. You can see exactly how to do it. So, just to recap here, you can build and deploy apps using Gemini 2. 5 Flash directly inside gemini. google. com. If you want to create code and then preview it, so for example, this p5 js game, then you would create the code directly inside AI studio and then go over to p5js. editor. org and plug that code in and start playing it. Right? So those are two different ways that we can use this. Additionally, what we can also do is we can go over to something like client. And by the way, if you're wondering, okay, how do you set this stuff up or where are all the links, etc. How do you get access to all this stuff? You can just get access to the full SAP on building and automating with Gemini directly inside the AI success lab link in the comments description. Basically shows you how to build anything and automate anything with Gemini 2. 5. And everything that I'm showing you today has been broken down into step-by-step tutorials and guides. So you can steal this stuff from me. All right, so let's go back into Visual Studio Code over here. And then we're going to run a new task. And inside our extension client, we're going to go to settings, open router, and we're going to select Gemini 2. 5 flash preview from the options. Now, client have actually officially confirmed that flash, the new model is directly inside here. And you can switch between different APIs as well directly inside the chat here. You can also use Google Gemini 2. 5 flash preview thinking mode here as well. And then we can hit save on that. And from here, we can build whatever we want. So for example, we say just something quick and basic to show you exactly what you can build. Let's say something like okay build out a and we'll take some examples from the AI profit boardroom link in the comments in description if you want to join this community with coaching et etc on AI but we can say here okay create beautifully designed snow calculator optimize for the keyword actual content on the page and relevant tool right we'll go inside there and then we'll also say right make this pink gradient fun emoji plus make it look sleek, modern, and beautiful. Plus add FAQs like then we can hit

enter. Make sure that you've got the right API selected, hit enter, and that's going to start coding out. What you can also do is inside Root Code, if you paste the same sort of prompt, you can go inside here and you can click the AI button and that will actually improve the prompt for you if you want to get better prompts or just make your prompts more detailed. But you see how it's basically instantly building this out, creating the code for us. And we really didn't have to create much work there at all. Jose Pinto says, "Good one. " And that's building this out. Now, we're going to hit run command. We'll open that up. We've got the SEO snow day calculator built out. Now, being 100% honest, the UI of this does not look very nice. So, what I would normally do is either I'm going to be very patient and go back and forth with Gemini Flash for ages or I can switch now to Anthropic and then inside Anthropic Claw 3. 5 Sonet and then inside here we can say okay make the UI more beautiful, sleek, modern, interesting. All right. And then Claude 3. 7 Sonnet is just a lot better for improving the prompts and improving the UI on the actual code that you build. All right. So, that's what I'd recommend. Now, we're going to close that because I've shown you how to build with client. You've seen Gemini. You know how to build with AI Studio. What's next? What we can also do is we can go over to NA10 and you can build out workflows with A10 using Flash as well. Okay. So, if we go inside NA10. io, this is a really powerful tool for building out automations. So, for example, like I've built out voice automation, Slack bots, all sorts of agents inside NAN. And if we go inside create workflow here, we'll just create a little basic example. In fact, what I can do is yeah, create workflow. Then we can put this as a chat trigger just to show you a really basic example what we can build. And then we're going to add an AI agent. Inside the AI agent module, what you want to do is just add chat model. And you could either use like Google Gemini chat model and create a new credential there. Or we can delete that. So let's close that. And then inside here, we could select open router. Open router. We can add a credential inside here. So let me show you how to get that. You're just going to go over to open router. Select keys. Create API key. Type it in. Copy that. Go over to NA10 again. Paste in the API key. Hit save. Excel off that. Select the model. And then you can use Gemini 2. 5 flash preview 20. And also thinking mode. If you be an agent and it doesn't require much thinking, you just want instant responses, then you should just use flash preview and you're good to go on that. So, if we want to test this out, I'm just going to put some system instructions inside the AI agent. So, let's say you're helpful assistant and we'll say, okay, is 100 * 40. It's running the AI agent. As you can see, cuz it's flash, it's really good for building out AI agents. And then if we open this up, it said, here's how to calculate 100* 40. Here's what you do. And then it's given us the answer right there. All right. So, that's how you can build AI agents as well. So if you want to just steal that example from me, I'll actually download the JSON and then inside the AI success lab link in the comments completely free. You can download that JSON from me and you can grab the example agent and all your agents that you use, you can just switch to flash if you want to as well just by changing the chat model that branches off as you can see. So just to recap here, we've talked about Gemini and building apps directly in there. We've talked about AI Studio and how to code with that. We've talked about Visual Studio Code and how to code with that. And we've also talked about how to build AI agents using NA10 as well. So you can see the power of this and how easy it is to build out almost anything that you want using Gemini in this process. Let's keep going now. So what we're going to do from here is we're going to start using something called Bolt DIY. BoltDIY is like a offline version of Bolt where you can code with it directly and you can also use Gemini Flash inside here. So if you want to set this up quickly, pretty simple SAP inside the AI success lab, which is completely free for you to get access. And if we go inside the instructions, we'll do a direct installation. Paste that in. Then we'll PMP install that. And literally all you're doing is just copy and paste in, right? It's pretty simple. And then we're just going to do that. Now, if we need to pull in the latest model from Bolt, we can just grab this, we can plug that into Claude and say how to set up. And if you have any issues, like you see over here, we can just go back in here and say, "Okay, help with this. "

And then we'll copy this. And what that's going to do is start running Bolt DIY locally. So, we can build with it. If we copy that local address, then we'll open that up in a new tab. Now you can see we can start coding with bolt. And if you select open router and then you can choose Gemini 2. 5 Pro preview here and just get access to 2. 5 Pro preview to start building stuff out. Right now you just need an API key. So you can get that from open router. Let's go to keys in the top right. Create API key. It's good practice is just give it a name you can actually remember. And then we'll copy that code. Go back into bolt. plug that in and say, "Okay, build a SEO calculator. " Right? Something like that. I'm just showing you basic examples for the video before anyone comments and saying, "This isn't that impressive," etc. And now that should start coding out. Now, the only issue you're going to have with this is sometimes this runs a little bit slowly depending on how fast your laptop is, but you can see it coding on the right hand side using Gemini 2. 5 Pro preview. But again, you would just switch to flash preview if you want. So, we can switch to flash preview, which is the latest model. And then we can say, okay, make the UI beautiful. And if we select that text and then click on the AI section under here, it will actually improve the prompt so that we get better outputs. So, let's wait for that to run. As you can see, then we're going to enter. And you can just improve it as you want, step by step. does take a little bit of back and forth, but honestly the outputs worked first time round, so that's good as well. And then if you ever want to shut this down, we could just quit the terminal, terminate that. We'll delete the API key, and we're good to go on that. Right. So, that's basically how you can build anything with Gemini 2. 5 Flash, the new versions. So, we've talked about client and Visual Studio Code. We talked about NA10, AI Studio, Gemini, etc. all the SPS inside the AI success lab if you want to get free access to this plus an awesome community of 6,700 members that you can ask any questions to. On top of that, if you want to get help, support, tips, etc. implementing this stuff, feel free to get the AI profit boardroom link in the comments in description. This basically focuses on just more practical use cases and supports you with trying to implement this stuff. It can get complicated. You're not sure how to use it. Maybe you need help and support actually implementing it and you're getting some bugs, etc. This is one of the best places to go to help you support you on your journey towards using AI, saving time, making money with AI, right? And you can also post inside the community. You can DM us. You can jump on the weekly live calls as you can see inside the calendar. And also inside the classroom, we have tons of training on all different sorts of agents and our crash course which is absolutely amazing for learning how we've actually implemented AI in our business to make money. So feel free to get that link in the comments description. Price is going up next month. So, make sure you lock in your lifetime deal now and I will see you on the next one. Cheers. Bye. New Gemini 2. 5 Pro, how to build anything. So, today what I'm going to show you is how to build basically anything in many different ways using the latest update that just came out 24 hours ago from Gemini 2. 5 Pro. And the coolest thing about this is actually free to use if you have the right techniques. I'll show you exactly how to do that. By the way, if you want all of the for the stuff that I show you today, you can get it inside the AI profit boardroom, just go down to Gemini Pro 2. 5 Pro down here, and you can get access to everything right here, including all these techniques that I'm about to show you. I'm going to show you nine different ways to build anything that you want using Gemini 2. 5 Pro. And we're going to kick it off with something interesting called Bolt DIY. Now, BoltDIY is a really power to start using AI to build stuff locally, right? And it's kind of like Bolt. But bolt new is a paid tool where you have to pay to build SAS or apps or that sort of thing. But you can actually use it for free directly inside Bolt Local. So, let me show you exactly how to set this up, how it works, etc. First thing that you're going to do is you're going to go to terminal, right? And then inside terminal, let me show you how to do this. By the way, if you want access to all this stuff, you want to learn how to do this stuff, get it inside the AI profit boardroom. I've got full SPs and all this stuff and how to do it right here. And we're just going to get straight into this. All right. So, if you want to start coding locally with Gemini 2. 5 Pro to build apps, games, tools, etc., then you can use this terminal command like you can see right here. And then we're going to mpm install like you can see. Just going to paste that in. Then from here, we're going to use this. Right. Next up, what we're going to do is mpm rum dev. That bad boy. All right. And then that's just going to start running this boom shaft kalaka. That was easy, wasn't it? Then what we're going to do is we're going to open this up. So this is running locally now. This is kind of like a free version of bolt new, right? So you don't need to pay for this. It's completely free to use. And if you select open router in the settings right

here, scroll down, you can get access to the latest version of Google Gemini 2. 5 experimental, right? And this is how you can start coding stuff, right? So you can switch to Google over here, but I can't see 2. 5 Pro experimental in the list. So the best way to use this start coding with the API key, which is going to connect Google and its AI to bolt new bolt. local. Then you just select Gemini Pro 2. 5 experimental for free down here. And then what we need to do from this part is start getting an API key and plugging it in. Right? So you see where it says open router API key. That's where we're going to get an API key and plug this in. Now if you want to get an API key, pretty simple stuff. So you're going to go to open router. ai and you can get a free API key and then use this for free. All right. So if you go to the key section over here, we can just click create a key. I'll do that off screen. And then from here we can start getting into the magic of coding with bolt local. Now even it for example sometimes with these APIs they don't work right? So you know if it's a beta or if it's just been released like 24 hours ago sometimes these don't work. But I'm going to plug in this API key. Hopefully it'll work. And if it doesn't I'll show you other ways to do this in a sec. Right? So if we say build a to-do list app inside here. All right. We're using Google Gemini Pro 2. 5 experimental. This is starting to build out our app that we asked it to develop. Right? And you see how fast it is, how easy it is to do using Bolt Local, sorry, Bolt DIY. And bear in mind, right, if we go onto Bolt. new. So that's the to-do list. We'll come to that in a second. But if we go to Bolt new, you have to pay, right? So if you go to my subscription here, if you want 10 million tokens, it's $20 a month. You want $26 million tokens, $50 a month. 120 million tokens, $200 a month, right? But if you want very similar functionality completely for free with a free API, then just set up Bolt DIY. And you've seen how I set it up in 2 seconds. If you want the instructions for setting that up, it's inside the AI profit boardroom link in the comments description. So from here, we can start building whatever we want, right? So for example, I've got a to-do list app right here. We type in whatever we want, right? Figure out Gem Line Pro. Bang it in. We got the list ready to go. Right? And then if you want to for example host this stuff right then what you can do is you can go back to code over here you download the code right so that's our folder our react folder and then what we can do from here with that code that we just created is we can actually host it and we can host it for free on a subdomain. So, if you want to host it for free on a subdomain, you're going to go to netlifi. com and then you're going to go from here and you're going to click on import an existing project, try netifi drop, and then just drag and drop that folder that you just downloaded from Bolt Local. Right? How easy is that? So, incredibly easy, not just to build your own app, not just to do it for free, not just to get free access to bold. local, but also to get this bad boy deployed on a subdomain. Pretty crazy stuff. And that literally just took us like 1 minute or so. So you can see how easy this is, how powerful it is, etc. And you can do it in the space of like literally minutes to get this stuff deployed and hosted and that sort of thing. If you wanted to build a game or app or whatever, if we go back to Bob DIY here, we can also switch between API keys. So if you want to test stuff out, there's loads of free stuff here. So if you go down the list, you'll see there's a lot of free ones. So for example, like Llama 3 AB preview is free. All these ones with brackets and then free, these are free API keys that you can just play with. But I would recommend Google Gemini 2. 5 Pro. It is probably the most powerful API key in the world and you can get access to it for free, right, for cod and stuff. So unbelievably powerful. Build the best SEO cost calculator you can with a sleek modern design. Wait for that to load. You can build tools, you can build apps, you can build games, calculators, etc., whatever you want. And it is super fast. I also find like this is a lot more reliable than other API keys that are free, right? So, for example, if you try and use the Deepseek 3. 1 API from Open Router, it's a little bit slow sometimes, right? Doesn't always do what you want it to do. It's a little bit temperamental. Whereas, for example, when you're using a process like this, it's simple. It's easy. It doesn't take a long time. And literally, in one prompt, you've got that ready to go. Look at that. Right. Here we go. Boom. Shaka. All right. And then we got the SEO cost calculator. You select your services. gives you the estimate monthly cost. Something that you would hire a team of developers to build and spend thousands of dollars, you can now do for free using Bolt DIY. And that is just one of the methods that I'm going to show you today. We've got many more to get through, my friends. And again, Saps are inside here. So, the next method that I'm going to show you is on client. Now, client is a free extension inside Visual Studio Code. All right? So, if you go to visualstudio. com

download this, right? And then what you're going to do is open this up in Visual Studio Code. That's going to open up something that looks like this. Then inside the extensions marketplace. So click over here, type in client. Allow that. And you can install I would recommend installing client and root code. These are two different ways to code. Root code is super powerful cuz you can actually use like computer use with it as well, which is pretty insane. And also client is really good as well. I like both of those extensions. I think they're great. And then inside the settings once you've got client installed. So you can see client on the left hand side here. Then you're going to go inside the settings and then you can input a Google Gemini key like you can see right here from the list. You will not find Gemini 2. 5. Right? So if you want to use Gemini 2. 5 instead what you want to do is go to open router and then inside open router we're going to put 2. 5 and we'll look for Google Gemini 2. 5. Let's have a look what we got here. Yeah, there you go. All right. So, you can just find it from the list right there. Let's have a look. Yeah, Gemini. It's called So, it's called Gemini 2. 5 Pro Experimental 03-25 colon free. So, it's free. It's available inside open router. You can use it inside client. And then, you know, you can use the same prompts you use inside Bolt. So, if we go back to bolt over here, I'm just going to shut this down because that's probably taking up a fair bit of memory in the old bolt. All right. So, we'll leave that for now. And that is terminated on the local host. But we can always go back to the old chats if we run it again. So what we can do from here is we can say now build an SEO cost calculator. So you can do very similar stuff inside client. But here's what I would say. It's like when you're using client, client is a lot slower for building stuff out inside client versus using bolt. local. So if you saw the way that I coded bolt. local previously on bolt, way faster whereas if you're using client or root code, it's going to take a bit more time. And I think it's because it's updating files locally. And I think as well because it's running it through the custom instructions that are set up in the settings as well. So you see how it's like writing the content for the page and it's creating all the content and text for the page as well. Analyzing stuff. I think Gemini Pro is probably your best bet. Some people are asking like can you use for example for trading and analyzing can you use Deep Seek 3. 1? I wouldn't recommend it cuz Deepseek 3. 1 is a base model not a reasoning model. What that means is it's not as good at thinking and reasoning and logic and coding as something as analytical as Gemini 2. 5 Pro. So I would rather use Gemini 2. 5 Pro for analyzing and that sort of thing. Right? So we're coding that out. We're building out websites using client. You can do exactly the same thing inside code as well. So if you switch to R code here, you know, this is method number three. You can do exactly the same thing. Again, you would select from the list. You can use deepseek if you want but again I would always prefer to use Gemini 2. 5 Pro and again you select it from the list hit save. Done. Right. The difference one of the biggest differences between Klein and Root Code is that Root Code is easier to install MCP servers. Right? So if you go inside the MCP section over here, it has this MCP server section and then you can see which MCP servers are running. Right? I'm just going to delete that one for now. But you can see we're using like Plexi ask right here. So we can connect RC code directly to Plexi. So the way they use MCP servers is very different, right? Whereas if we go over to client, they actually have an MCP servers marketplace. And then you can use the Gemini 2. 5 Pro API to install MCP servers such as connecting it to stuff like Google Calendar, Discord. I've connected it to Plexi. OAM is another good one. There's lots of good stuff right there. And this is completely free, right? So just to recap here, we've talked about three different methods. All right, we've talked about using bolt local, which is free and you can build anything using that. You can also, for example, use client brew code. These are many different options. All right, let's keep going down the list cuz we got a lot to get through here, peeps. All right, so this is something that I'm going to recommend called make. com. Make. com is really good for like linear workflows. All right. So, you can build like AI agents using make. com. And I'm just going to grab a password here. All right. We'll plug that in. And we'll log in to make. com. Wait for that to load. And what we can do is build like AI agents with Gemini 2. 5 Pro. All right. So, let me show you an example of what we built yesterday using similar sort of processes. So, for example, we've got where have we got it here? So, we've got tools like this. This is actually made with DeepSeek, but we can duplicate it. So, I'm going to clone this right now. Call this Gemini 2. 5 Content Creator. Copy. All right. Hit save. And then what we're doing is we're taking keywords from a spreadsheet, plugging that for Open Router, sending to WordPress. All right. Now, inside the

settings for Open Router here, we have the prompt for generating the content. We have this connected to Open Router. And then we're just going to scroll down here. And instead of using Deep Sea V3, I'm going to switch this over to Gemini 2. 5 Pro Experimental. Right there, you can see it. We're going to hit save. So again, you would just add your open router connection, your WordPress connection, and your Google Sheet connection insidemate. com. And you can build like oneclick AI agents using this process as well. So this is method number four for using Gemini 2. 5 Pro. And then for example, if we choose where to start over here, we're going to choose manually from our spreadsheet of keywords. This is an AI SEO content writer. You can use this for whatever you want, right? Like you could take keywords and then plug it through open router and then do tweets or LinkedIn content. You could automate emails. You could fire off emails to potential prospects, whatever you want. Okay, we're going to hit save on that and then we're going to hit run once and then we can start using this process. Now, let's have a look here. What's going on here? Why have we not why is this not working? I'm just going to switch my open router connection. Make sure we're using the right model. We'll test that again. There we go. All right. So, that's working nicely. Now, if you want to see an example of the output, think of this is because the other API connection that I had inside open router was the old connection and I've actually deleted the API key since then. So, if we log into my WordPress over here, whilst we're waiting for that to load, I'll show you some examples of what I did yesterday using the exact same process. So, here's a Gemini Pro experimental example. We're going to click on preview. We've got the content and then you can see it's just created like a nicely formatted blog. Reads pretty nicely. It's interesting. It's all targeted to the keyword we wanted to write SEO content about. And this was free. Right? Make. com is free for a thousand operations. Google Sheets is free to use. WordPress Open is free to use. And it's using the latest API keys from Gemini. Right. Now, if we take that and we'll go over to Z GPT, we'll plug that in and we'll just check the AI detectability on that. Is it AI detectable? No. Zero GPT, which is an AI detector is free and it is detecting this is 100% 98% human written. It says your text is human written, right? So, automated content. It's all customized to us, nicely formatted, doesn't cost us anything. We did it in one click and it all links out to our personalized funnels like you see right here. Right? Pretty amazing stuff. Plus, it can publish automatically for us as well if you want. So, that's how the whole process works. You can see it worked quite cleanly there. And if you want my best automations for that, go inside the AI profit boardroom. We've got like a whole module on make. com automations and there's 12 different templates you can get inside there. So, make sure you get that stuff. All right. So just to recap, we've talked about BoltDIY, Klein, Rode, we've talked about make. com. You can also just go directly inside the chat, right? So if you go to a studio. google. com and then you select gemini 2. 5 Pro Experimental over here, create a prompt. You can just build whatever you want inside here. Now, if you want to see an example of that, let me show you an example right here. So I'm just going to go through the library. What else we got? Yeah. So I created like a pixelated dinosaur game yesterday, right? So here you can see the code. I just went to Gemini 2. 5 Pro. I said, "Make me a captivating endless runner game blah blah. Do it in 5JS. " It gave us a code right here. We can copy that. So we'll copy that to clipboard. And then it even gives us the instructions on how to get this set up. So if we scroll up here, we're going to go to editor. p5js. org. Let's go to that website. Copy that. We'll paste in the code. We'll hit play. Boom. Game built. Right. Pixelated dinosaur game. Live in the dream, sunshine. All right. And you can now build games as well for free with Gemini 2. 5 Pro. So that's another method as well. Like you got so many different options right here. Choose the one that works the best for you. Depends on what you're building as well, right? So for example, if you just want to build a simple game, then you're probably going to use something like AI Studio. If you need to build like a really big website, you're probably going to use client or root code for that. If you want to build out a SAS app like super fast, then all you need to do is you can just go over to Bolt Local and just build that out for free, right? So many different options depending on what you want. If you have a linear automation like this and you need to post from one platform to another one, then make. com is really good for that. Next up, what I want to show you is browser agent web UI. All right. Now, I've actually set this up with Deep Seek. The instructions are inside the AI profit boardroom, but you don't need to use Deep Seek. You can use Gemini 2. 5 Pro. And then I'm going to show you an example video tutorial of this so you can see exactly how it runs. I'm not going to show it live on this video cuz it takes about 15 to 20

minutes to set up. All right, but you can basically set up browser web UI like you can see right here. You would plug in the LM provider as Google, not as O Lama, right? And then like you can see right here, it can just go off and start doing stuff for you. All right, so check out that video. It's on YouTube. You can see how the AI analyzes the page. The good thing about Google's Gemini model as well is it's a vision based model, so it can do its magic just like that. And again, instructions inside the AI profit boardroom. Next up, what I want to show you is you can use deep research agents, right? So if you want to build like really in-depth research reports, you can go to deep research inside Gemini. This is another free feature as well. It's not necessarily using Gemini 2. 5 Pro, but if you need like the best research, better than a human, faster than a human, this is how you do it, right? So for example, if we say, okay, research the latest chat GBT40 image update, it should start working its magic. And the cool thing about this is you got everything inside one ecosystem, right? So, for example, you need to do deep research, boom, there you go. You need to do something like personalization based on your search history. Boom, there you go. If you want to build out some tasks with like maybe a website or a SAS or a landing page, then just go to 2. 0 Flash and use the canvas feature. Like everything is what in one side, one ecosystem, one place. And that's the beauty of this is you're not going from place to place and messing around like that. So, if we hit start research like so, we can build out these deep research AI agents that can just go off search the internet. We can leave that tab running in the background as well if it's taking too long for you. And then we can just go off and do that. All right. The final two methods I want to show you cuz I've shown you all these different methods already. Number one is using NA10, right? So if we go over to NA10 over here, you can actually create AI agents directly inside NA10. And these are kind of like oneclick chat bots or workflows where you can connect different apps to each other. Difference between say something like NA10 and make. com is that NA10 is not linear, right? And you can build your own chat bots inside it. Whereas make. com is very like you go from point A to point B, boom, straight line, right? So for example, let's have a look at this workflow that I actually built out yesterday. If we have a look at this one, I'm just going to check the open rout to chat model and we're going to select instead over here Gemini 2. 5. Let's have a look here. Put 2. 5 Pro. Yeah, there we go. So, you're going to select Gemini 2. 5 Pro Expand Experimental. You see, right inside the AI model and we have an AI agent where you can essentially like automate sending emails or going back and forth inside the chat. You can also add a memory in here as well. So you can add a memory that allows it to remember previous messages so that it remembers stuff that's going on. If you want to publish this, you can also make this chat publicly available. Grab that URL, publish the workflow, and you've got a chatbot that you can share with the world or embed to your website, etc. And then if we go inside the chat here, we'll say, "All right, we send a message to me at julianoldie. com telling there's new ways to build anything you want with Gemini 2. 5 Pro Experimental. And then I'll list out all these different methods here using this. Keep it fun plus conversation. " And then from there, we can start building this out. The other way that we can do this, we can delete the open router chat model if the API key doesn't work. I told you it' be temperamental sometimes. Right. So if that's if you're having any problems with that, you can add inside here. Let's X off that. What are you doing, mate? All right. Click on chat model over here. Click on Google Gemini chat model. And then you can select 2. 5 Pro Experimental. Give you the same chat message. And now it's using Google Gemini to generate that response like you can see. So the chat model is working. Actually what's interesting is you go inside Visual Studio Code and client and open router works perfectly with Gemini. Whereas for example if you go over to NA10 and try and do the same thing. It's better to have a Google Gemini chat head from what I can see because that API key worked perfectly first time round. Right. So we're good to go on that. I'm going to hit save on that. If you want to download this template from me, if you want to just steal this stuff from me, I'm going to download that and I'll plug it into this section right here. So, add a new resource file. And if you want to just import this workflow, it's waiting for you right there. And it's an example, we'll hit save. Boom. There you go. All right. And then if we check our emails in the send box, here we go. Look at that. Send it right there. The only thing that I would say here is make sure you write it in HTML and make sure the output is in HTML as well because then the formatting is going to be much nicer. You get the point, right? If I say write it and send it in HTML

hopefully the formatting will be much better this time. So, let's wait for that to load. And again, like you can connect this to any tool. I'm just giving you a quick example for the video. If you want to see all my best templates for this stuff, type in NA10 into the AI profit boardroom search feature here. And then you'll see all these NAN Asian workflows. Like you see all these different resources here. We've got loads of different AI agents. Like you can see here as well. NA10 sales agents. Like it's not limited to just sending emails. It's just for the sake of this video that I do that. Right. So if we go back now to NAN should have sent. Let's double check the send messages here. Yeah, there we go. Look at that. Beautiful. Couldn't have written that title better myself. That's a perfect one for YouTube, right? That's ready to go. And then you can see it's just written it nicely formatted. So, just make sure you output it in HTML inside the prompt settings. And then it's standard magic, isn't it? And then we've got the deep research report ready to go. Back enable ready there. So, that's pretty cool. All right. So, this is the deep research report we were talking about earlier. It takes about 5 10 minutes to go and then we're good to go on that. And the final method that I want to show you, all right, in a long video, this one been through a lot of different methods to build anything with Gemini 2. 5 Pro. So, what you going to do with this app right here is you can select inside the models Gemini 2. 5 Pro Experimental. And then we can say build me a calculator app like you can see. We'll hit enter over there. and it will just build out the app. Kind of similar to Bolt Local, but it's not going to be as big a product with as many different features, right? And it might take a little bit longer, but yeah, you can see how you can just get started with this. The good thing about this is like you don't need to use a terminal to just get this all set up and you can turn your idea into an app for free using this process. How easy was that? So, there you go. You got the calculator. It's a better designed calculator than Windows used to have a few years ago, to be honest with you. And you just did it with AI. Boom. Right. If you open up the sandbox as well, we can open this up and we've got the code ready to go. There we go. Right. So, we've talked about nine different ways to build anything that you want with Gemini 2. 5 Pro. All the SPS are inside the ARI profit boardroom, including all the instructions, NA10 agent workflows, etc. And the prices for this are going up at the end of the month. So, make sure you sign up now before you miss out, my friends. All right, this community is just focused on one thing, which is making more money and saving time with AI. So, if you want to do that, feel free to get it inside the AI profit boardroom. And if you want to jump on weekly Q& A with us, you can jump on the live course right here. Ask us any questions you have. If you want to get all my best courses, SAPs, workflows, etc., it's all inside here. And inside the community as well, we've got 557 members, all focused just like you are on making more money and saving time with AI. So, if you post inside the community, you can ask any questions you have. You can get help from all the community members here. And we're all here just focus on one thing, which is rising tide lift boats. Basically, let's help each other out and let's learn together and make sure we all don't get replaced with AI. All right? And if you want to get a free one-to-one SEO strategy session, that's completely free. We'll show you how we take websites from zero to 145,000 visits a month and generate hundreds of thousands of dollars in sales on autopilot. This free link building acceleration session will give you a custom tailored link building game plan to make more leads, sales, and profits from your website. You'll discover the secrets building right answer any questions you have and you'll learn 10x traffic based on what's working for us and our happy clients like you can see right here right feel free to get that link in the comments description completely free for you my friends Gemini 2. 5 the new update as you can see right here versus Claude versus Chat GBT who wins we're going to be comparing them side by side and testing which ones are the best got a bunch of different tests and use cases to check this through. Let's see how it goes. All right. First thing that we're going to do is we're going to test this prompt right here for writing content. All right. So, we're going to say write a 2,000word article. The feature of AI agents, the impact on automation, etc. And we'll hit this into each option like you can see right here. All right. So, we're going to be testing this out versus each of these models. Use chat GPT4 cuz that's kind of like the default one for writing inside chat GPT. And then we'll see which one comes back with the outputs and performs the best. First thing I would say is Grock has come back to us and seems to be the fastest. But actually, if you look at the content itself, look at those big walls of text. Look at the way it's written, right? Would anyone actually read that block? And so for me, that's one of the worst that I've seen so far. Let's have a look what we got here from chat GBT40. So this is pretty nice the way it's written. So what is an AI agent? It's dumb down the language, made it nicely formatted, made it really easy to read as you can see right here, and that's performing pretty well. I'm playing English, an AI agent and software that takes action for you based on goals and prompts. You an objective, it decides what to do next. Pretty nicely written. Feels very humanized. It's still writing out, but you can see how fast it is. And it's also doing like the 2,000 words you

ask it to. So that is doing a great job. Let's have a look at Claude. Claude is one of my favorite ones if you give it the right prompt. But a prompt like this, look at this. Just really long, boring walls of text. I'm going to put that as coming in third. I think maybe slightly better than Grock, but not as good as Chat GP. And then we got Gemini's written the blog here, but that is more like a deep research report, right? Who would read that? And so out of all them, I'm going to say that Chat GT is winning by a long way, right? Long, long way. So when it comes to writing, pretty good. Now, one thing that I would just note here is if you give it the right prompt, I've got this prompt inside the AI profit boardroom here. If you give it a better prompt, Claude is actually pretty good for writing. But it seems like if you don't have the right prompts in place, look at how bad that is compared to the output from chat GPT that required basically no prompting at all. Huge difference right there. All right. So, next up, what I'm going to do now is I'm going to go into each and I'm going to say, "Okay, new chat, create a beautiful landing page for my SEO agency's website. " All right. Now, one thing to note here is if you want to use Gemini 2. 5 Pro for free, you can do that inside AI Studio. Go to create prompt, then select Gemini 2. 5 Pro, and you can get access to this for free. All right. So, you know, that's one of the other options I would say here is all of these models are free up to a limit, but actually if you use AI Studio, that doesn't seem to have a limit. Like, you can use it for free as much as you want, peeps. All right. So, we're going to go back here now. We'll do the same inside Grock as well. The cool thing about Claude is it has a canvas option. So does chat GPT. And let's go back here. What we're actually going to do is use 2. 0 Oh, flash for generating this inside Google cuz that's a canvas option. We'll hit enter and we'll see what we get back in a second. All right, so now we've got the code back from chat GPT. Let's run this. See how it looks. It's not bad. The content itself actually reads really well, right? That's better written than most people. It's got a nice little CCA to put in your website URL, nicely designed icons. I don't like the front end so much cuz it's super boring in terms of the way that it's done. Claude is still writing. Brock didn't even figure out we need HTML. So that's gonna finish last again for me to be honest. It's been a brutal test for Grock already to be honest with you. And we're going to keep going through this. So you can see the output from Grock here. It said I can't directly generate HTML or CSS, which is mad because Grock can, right? It genuinely can, but it's refusing to. So Grock has to come last on that test. It's just no good. I don't know why, but Gemini is written in Chinese characters. The design is pretty nice, but for some reason, it's written it all in Chinese. Why? Why have you done that? I've never seen that before. All right, let's go back to chat. GBT is winning just by default for writing in English. But look at that from Claude. Claude has created the best so far. Nicely designed, looks really cool, good UI. I like the color gradients on it. Nice copy as well. So, dominate search results with datadriven SEO. Really nice headline. So, in this order, here's what I'm going to say. Claude has won that by a long way. Chat GPT comes in second. Not bad at all. It just could do a bit of color on the page, but you could always go back in the chat right there. Gemini totally failed and gave us something in Chinese, and I still don't understand why. That's a bit mad. And then Grock said, "I can't create HTML for you. " Even though it's got a preview section inside Grock 3 for HTML, it just didn't understand. Right? So, out of all these, he's going to go Claude, I would say, for doing this stuff, which is crazy. I don't understand. Like, Grock seems to have got worse or something like that. So, let's keep going now. And next up, what we're going to do is we're actually going to use a prompt from the AI profit boardroom for creating games. All right. So, we're going to go inside the classroom here. And I've already done this inside AI Studio. So, I don't need to do this with Gemini 2. 5 Pro yet, but I will do it with the rest. Okay. So, we're going to go inside Claude here and I'm going to say, "Make me a captivating endless runner again. Key instructions on screen. Use P5 JSON HTML. Pixelated dinosaur and interesting backgrounds. " Right. We'll use the same inside Chia Chiba T40 Grock. I just I don't know what it's going to come back to me with. Oh, yeah. Look, it's actually doing its magic now. That is surprising. And then if you want to see the output from Gemini 2. 5 Pro that I've already done. Let me come back to this right here. So this is the prompt that I used yesterday. We'll scroll down and we will get this code and I'll show you what we got back from Gemini 2. 5 Pro. All right. So if you want to preview

code like this, if you're using the same prompts as me, you can plug this into editor. p5js. org. And then inside here, we can hit play. And you can see that we've got this bad boy ready to go. All right. Looks pretty cool. Nice little game. Functions perfectly. Quite addictive. And we've got the score racking up in the top right over there. And then it works perfectly. If we want to restart, we press enter. Press space to start here. Boom. Beautiful. All right. Really good from Google Gemini. So, they've done a great job there. Claude is still coming back to us. Chat GPT has come back now. So, let's see what we got back from chat GPT. We'll copy the code and we'll test this one. So, this is the code from chat GPT. Let's see what we got back. What is that UI? I can't jump or anything. It says, yeah, that's pretty poor. That is not a dinosaur, mate. Not a pixelated dinosaur and definitely not a dinosaur. Claude is still coding it out. So, wait for that. And we have the code back from Grock 3 as well. So, let's see what we get back here. So far, Gemini is winning by a long way. The UI is not quite as nice here. You see how it's green on green, but it works. It's definitely better than chat GBT's option. It's better. It's not quite as fast and as smooth. If you look at the UI of this versus the UI of Gemini, this looks a lot nicer to be fair to it. So far, what I would say is that Gro is coming in second. So far, Gemini is winning. Chat GBT, I didn't like the UI at all. Let's have a look what we got back from Claude. Is it opening up? It's giving us a preview, but there's nothing inside the preview, which is strange. All right, so we're just going to copy this code, and we'll go to p5 js and just preview this again. Let's see how Claude performed. Pixelated Dino Runner. Let's test it out. Not bad, actually. That is a hard game to play, though. Look at that. How are you meant to That's hard, mate. Yeah, I like it though. Not going to lie to you. So, I would say if it was actually beatable, cuz it seems like it's almost impossible to beat that, then I would probably put that number one or joint first. I would say that Claude and Gemini did a fantastic job there. Gro's output was okay. Mediocre. I'm going to put that at number three. Chat GPT's UI pretty terrible. It's just a block with a bit of text, right? Not great. So you can see who's winning there and who won the game right there. All right. So let's keep going through now. So we've tested it for content writing, for coding out landing pages. We've tested each of these models for creating content and also building out games as well. What are we going to do next? So the next one that we're going to do is this is going to be kind of like a reasoning test. So, we're going to go inside Claude and reset for each one here. Switch back to 2. 5 Pro experimental inside Google and we say there's a tree on the other side of a river in winter. How can I pick an apple? It's a reasoning test. And what we're trying to do here is figure out, okay, which one is the best for reasoning logic? Which one gives us the best solutions? And which one most intelligently looks up problem and then gives us solutions, right? So, we're just waiting for each one to come back to us right now and we'll see who performs best. So let's see what we got here. All right. So it's recognized that the biggest challenge is apples and winter. So obviously in winter if you're trying to pick trees on the other side of a river, it's not going to be ideal, right? Because quite likely even if you cross the river, there's going to be no apples in winter. So it's figured that out. It's analyzed the situation really well and it's given us a bunch of solutions right here. Pretty good. Let's check out now. Claude has recognized the same, but it doesn't analyze it in the same level of depth, right? So for example, if you compare Gemini 2. 5 Pro, he's really thought about it. He's given us a really in-depth answer. It's analyzed all the problems and solutions and the conclusion, etc. I would say Claude doesn't do that in as much detail. If we have a look at Grock, it's recognized typically don't grow in winter. But I don't like the way the answer is not logically set out. Right? So if you look at you've got these titles, it's really easy to read. I honestly I think that Grock has been tuned down in terms of its reasoning. Like it just doesn't seem to be as powerful as it was when it was first released. I would say that is harder to read than this one right here. And it's just not as logical as well reasoned out. And let's have a look at chat GPT. So let's break this down. It's an apple. There's a tree. It's across a river. It's winter. So let's look through it step by step. But the biggest problem here is it's not figured out that there's actually no tree. There's no apples in winter on the tree, right? It's not figured that out when it's done all this analyzing. So, if I had to put these in order, I'm going to go with

Gemini, then Claude, then Gro 3, and Chat GBT40 comes last. Now, if we wanted to give this a fairer way of comparing each model, then what I would actually recommend doing is you switch to 01 and use one of the reasoning models. And to be honest, if we did that, we probably get much better outputs. But I believe doesn't chat GPTO. Let me have a look at this. No. So, yeah, if you wanted to switch this over to, for example, 03 mini, let's test it out now and we'll see if that figures out. So, we're using chat cheapy 03 mini. He's going to use his remod reasoning model to figure this out. That's even worse, isn't it? Look at the quality and responses of chat cheapy 03 mini versus Gemini. Like it it's totally night and day. So let's have a look here. Says worried about crossing a cold icy river in winter. Wondering blah blah. And it still doesn't recognize that there's no apples on a tree in winter. All right. So chat cheap I mean using the most powerful reason model it has is not as powerful as Gemini 2. 5 advanced right so again chat cheapy comes in last whether you're using GT40 or whether you're using mini still down my friends all right so you can see the difference between the models how well they perform etc the other thing is as well inside chat GPT40 we're going to test this out now this is a really interesting test so we can do step by step we're going to do a test I actually used this morning using chat and the new image generation method. All right. Now, Claude automatically fails on this test or isn't even in the race here because it cannot generate images. So, you can't even count Claude at this point, right? But what we can do is we can test each of these models and chat is given me a really good option right here for image generation. All right. So what we're going to do inside Grock is we're going to use exactly the same prompt we use inside chat GPT and we said create a landscape YouTube thumbnail like this but it should be regarding the end of designers about new chat update that means we don't need designers anymore and I'm going to plug this in across each of the models right so the output from chat was top class I've never seen anything as good as this from any other AI model it was unbelievable really powerful just from the newest update from chat cheapy let's do the same thing inside gro All right. So, if we go inside Grock here, and we'll just we'll start a new chat just to keep things fresh. We'll grab that image that we use inside the chat GBT prompt. We're going to use that inside Gemini. Obviously, Claude is out the race and we'll see how it does. All right, this one might take a minute or so whilst we're waiting. And from what I can see, the output from Gemini 2. 5, not very good. Not very good at all. If we compare what we should have got versus what actually came back, not that great. If we use maybe we use like the actual main version 2. 0 flash. Let's test that again. So I'm going to re retest this. Grock as well is not done anything near as good as chat GPT. Right? Look at the output from chat GBT versus the output from Grock. This is I can't use that. You know what I mean? Can't use that. So you can see the differences here. And you can see that Grock is nowhere near the same level as chat GBT. Let's have a look at Gemini and we'll use 2. 0 Flash instead. So, we'll use exactly the same prompt. Let's grab the image over here. Plug that into Gemini and then we'll do the same thing and see what we get back. Yeah, look at that. It's just come back with a text message. Where's the image, mate? So, not great there. Yeah, if you compare all the models here, 2. 5 Pro's response was pretty poor. Grock didn't create anything near the same level of quality and Claude obviously can't do that at all. So, hands down, Chachi PT delivers like designer level quality responses, right? Like I can just download that, boom, upload it to YouTube, it's over. It's game over. You don't need a designer, right? That's the biggest difference here. And I think whatever you use for images, cheapo is like the new standard for all this stuff. So, thanks so much for watching. If you want to get access to my best prompts, tips, workflows, etc., including nine ways to build anything that you want with Gemini, feel free to get that with the SPS inside the AI profit boardroom. This comes with all my best AI workflows, tips, tutorials, templates, etc., and even a crash course on how to use AI and save time and make more money with it along my best SAPs. And also inside the Q& A section, you can jump on the weekly Q& A here and ask any questions you have. On top of that, inside the community, you can ask any questions that you want. We've got 557 members, all dedicated to making more money and saving time with AI, so feel free to get that link in the comments description. Prices are actually going up this month, so make sure you lock in your legacy price now

before you miss out. And if you want to get a free one-to-one SEO strategy session, feel free to get that link in the description. We'll show you how we take websites from zero to 145,000 visits this month and generate hundreds of thousands of dollars in sales on autopilot. And on this free link building acceleration session, you'll get a custom tailored game plan to generate more leads, sales, and profits from your website. You'll discover the secrets SEO link building, answer any questions you have onetoone, and you'll learn the best link building strategies for you, plus how to turn X your traffic based on what's working for us and happy clients like you can see right here. Feel free to get that link in the I appreciate you watching and I will see you on the next one. Thank you. The new Google Gemini 2. 5 Pro update just released at Kraken and it is absolutely free. All right, so what I'm going to show you now is how to build a game where we can just drive a car around and you know there's all sorts of crazy examples. I'll show you how to build like a full chess game and all sorts of crazy stuff in a second. You know, we can drive around. We can do whatever we want. We can go through this little train bit here. We can build crazy stuff, go backwards and forwards. All right. And it's really easy to build this stuff out, right? I think it's just bugged out completely there, but you get the point. So, let me show you exactly how to build it. And this is based on the new update that just came out like a couple of days ago from Gemini 2. 5 Pro. I think honestly most people don't understand how powerful this is and how much of a game changer it is, right? It's totally changed everything because not only is this like an awesome thinking model, not only is it beating, for example, powerful AI models like Claude in its benchmarks, but it's also incredibly powerful for like building out coding and taking an idea that you have in your head and turning it into reality, right? So, you can see it in terms of the benchmarks here. It's pretty much outperforming all the main models for most of the benchmarks here. So for example like humanity's last exam or Amy maths test is outperforming some of the most powerful models in the world but at the same time it is completely free to use right so if you want to go to a studio. google. com you can click on create prompt and then from here you can just type in your prompt and whatever you want to do. The other way that you can do this is you can grab an API key like you see right here. Then we can navigate to something like Visual Studio Code. And inside Visual Studio Code, we can start coding with the API inside extensions like client and ruode. Right? So if we go to the settings over here, we can select to open router Google Gemini 2. 5 Pro Experimental. And it's free. Doesn't cost you anything, but it's the best AI model in the world. How is that possible? And yeah, you can use it completely free inside a studio. The other thing to note here is if you go into gemini. google google. com/app. That's going to be paid, right? So, you have to pay for the model inside Gemini Advanced if you want to use 2. 5 Pro Experimental. Don't do that. Instead, do this. Go to AI Studio, go to create prompt, and then in the model section over here, go to Gemini 2. 5 Pro Experimental, and you can select whatever you want, right? And it is literally breaking the internet. Let me show you some examples. Right? So, a lot of people are saying this is literally the best model for code. So, it's extremely powerful. 1 million token context doesn't just agree with you. Flash is a genius, etc. Great tweet from McKay Wriggsley. You can also see some of the other stuff people are building. So, like this dinosaur game, I actually build that as well. And I'll show you exactly how to do it in a sec. And you know, people even building like crazy plane flight simulators like you can see right here. And it's not just limited to coding. It can do so much more. But that just gives you an idea of how powerful this is. And wow, my video just popped up right there as well. That's what we like. Julian Goldie rep in here. All right. So, let me show you some other cool stuff. So, for example, Punam Sony, she did this awesome tweet on like the comparisons. So, you got Gemini 2. 5 Pro over here versus GPT01 Pro. Like, which one is doing a better job? For sure, if you look at the realism here, I'm going to go with Gemini 2. 5 Pro, right? Like the physics and everything else look 10 times nicer versus GP01 Pro, right? And that's the difference here. It's like one of them's just better at coding, one of them creates better options, right? this car game we actually just created and I'm going to run you through like some of the best prompts and how you can build this stuff out right now. Okay. So, let me show you some examples of what we built. So, if you want to build the 3D car simulator, what you're going to do, simple stuff. You just say, can you create a simple 3D car simulator with 3. js and single HTML? Add clouds, mountains, roads, some trees, and a train going around. Right? Make sure it works on mobile. It's going to use its thinking mode. Like you see right here, we have 2. 5 Pro experimental figured out. And then once you've done that, you're going to see all of this right here. Okay. Now, this is actually coded in HTML. So, all you're going to do is I mean there's two options here. You can download it and then you can actually grab that code, drag it in here, and start using it like so. Or the other option is you can go over to something like LiveWeave, as you can see right

here, and start using it that way. And if it ever breaks or anything like that, just refresh the page and you're ready to go on that. Okay, pretty amazing stuff. Now, it's probably not going to be like perfect first time around, but you get the point. Like, it's incredibly easy to do. It looks pretty cool. Graphics are super nice, and it did it in one shot as well, right? Like, there's no messing around here. Like, it just it literally did it. Let's see if we can crash into that. All right, there we go. And then if you want to keep building out the rest of the game, you can just go back inside a studio. google. com and start doing this magic. All right, so that's how to build prompt one. I'll include all the prompts inside the AI profit boardroom if you want to get that link inside the description if you want to use this stuff. But yeah, incredibly easy, very fast. So let me show you how easy and fast it is actually to generate this. So if we take that exact same prompt and we'll start a new chat now. So we'll create prompt. Then we're going to click on this and we shall say okay, same thing, right? I like to press command and return just to send that quickly over. You've got a timer here. So, it shows you how long it takes to respond, right? So, if you actually open up the thinking, it's thinking out the whole process, etc. Looking at how long it takes, and you can see it's using this magic right there. So, it's going to plan out and reason this. This is not a base model. So, for example, something like DeepSync version 3, the new version, that is a base model, which means it's not designed for reasoning, right? Whereas if you use for example something like this, this is a reason model which means it's much better for coding, much better for maths etc. Right? And you can see how quick it's generating that code. Like it's incredibly fast. So it blasts it out. I mean it's going to be quite a lot of code but it pretty much codes it perfectly. And the good thing about this is you don't even need to be able to code or understand this stuff. All you need to be able to do is just download the HTML file and then drag it into a new tab. Right? Like that's pretty much it. So, it's still coding out right now. Let's see what we're up to. Finished off. All right. So, that took 66 seconds to generate this game. Now, imagine if someone was coding that out. How long would this take? Absolutely. To think about it and someone very skilled as well. Right now, what you can do is you can click the download option here. Got the download ready to go. And then we'll drag it in. And we have the game ready to load. Right. Pretty simple stuff. So if you want to build other stuff, here's another example. So this is fully working chess game in HTML. All right. So we ask it to generate a fully working chess game. Now, if we go back to liveweave. com, we'll delete that and then we'll plug this in. And we have the chess game. All right, chess game ready to go. So the code is in the top left. All we need to do is copy and paste HTML, nothing else. And then let's test this works. So can we play? It looks like we can, right? So, I can't move as a black cuz it's a white's turn. So, let's try that out again now. There we go. And it gives you the move history as well on the right hand side. Again, this was coded out in one single prompt. Absolutely wild. So, we just said make a fully working chess game in HTML. It's like here you go. Just go off and use it, right? No messing around, no back and forth prompting or anything like that. So, if we move the black pawn over here, we can decide to move forward or we can move to the left hand side. And yeah, incredibly easy. We can also move that and it works perfectly, right? How cool is that? How amazing is that? So, there's all sorts of crazy stuff you can build with this. This is just a start. Let me show you some other stuff. So, you can see here, this was an awesome tweet from Petri. I don't want to even try and pronounce his surname, but awesome tweet. Make sure you follow him on Twitter. And you can see he built like this Space Invaders game. All right, so let's try and test the same prompt out now. And I actually did that earlier. as you can see right here. So, I copied and pasted the same prompt from Petri like you can see and really simple and easy. Coded out as HTML again. All right. And then we're going to take that code, copy that over. We'll go back to LiveWave. All right. We'll paste that in. Let's see if this works. So, we've got the Space Invader controls at the bottom. And then you can see here now it's working perfectly. All right. And then I think we want to collect these health scores. So yeah. We just collect the health scores. And then we're having a jolly good time here, peeps. All right. I'm live streaming and coding and building out games. And it's effortless. And I didn't spend like 20 years learning how to code either. All right. So that's how cool this is, how powerful it is, etc. And we get little speed boosts as well. and it works perfectly. Now, I honestly believe the Gemini 2. 5 is the most powerful in the world for coding. I think it's going to be very hard to beat this, right? Like considering it's free as well and considering you can get an

API and start coding with it. The only thing that would make it better is of course like if you could open source it, but Google never going to do that. Like that's the secret source, I guess. I don't see Google We ever open sourcing this model, but it's super powerful stuff. All right. And yeah, you can also, for example, go into Visual Studio Code. Let me show you how to use this. So, you want to download a visual download Visual Studio Code at code. visisualstudio. com. Then from there, once you've installed it, go to extensions. And then from there, I think client is really good. client of Root Code. Pretty good. Root Code is one of my favorites, too. So, if we select the Root Code icon on the left once you've installed it, then we're going to go to settings and select Google Gemini Pro Experimental. Right? Now, the way to do this is two different options. You can number one, insert your Google Gemini API key, right? So, you go to a studio. google. com, click on get API key, and then from here, create API key, and you can grab an API key right there. Right? completely free to use. So you can grab one like that or you can do is you can go back to here and you can select open router. Right now if you select open router inside the API key section here you just go to openrooout. ai. Let me show you how to do that. There you go. And then once you've done that select keys at the top right once you've signed in and created an account click create key and you'll be able to use the Gemini model. All right. So if we go back with our API key, we would plug it in here. Select Google Gemini. And you see how it says colon free here. That means it's free. Free to use. All right. So once you've done that, we can hit save. We can hit done. This is a multimodal model as well. So for example, that means that if we have something in mind that we want to create, then we can easily do it. So for example, let's go back to live. If we go back to live reef here with the HTML from the Space Invaders game. All right, we could take that, grab that image, copy it here, go back into Visual Studio Code, paste in that image that we just created, and say, "Okay, clone this game and recreate it in a simple HTML file. " Right? Now, the benefit of this is that when you're doing this, and again, you see how there's some errors with the API key. So, if there's some errors like that, you can just go back here, select Google, and then select Google Gemini. But you can see it's going off and doing its magic, right? So, the way that you can do this, and we'll hit we'll allow that. We'll hit save. There is that when you're doing this, like you can save it locally. That means that if when you reopen Visual Studio Code, you've got the code ready to save locally. So that's the benefit of using this instead of like Google AI studio which is going to save everything on your account but you don't have it like to re-edit later and that sort of thing be a lot harder to find that. So that's basically how to build and create anything that you want. You know the new update from Gemini 2. 5 Pro from what I've seen for coding specifically is better than any other AI model out there and it doesn't cost you a penny. But yeah, it's an absolute gamecher. So if you want all the prompts from today and the examples etc. I'll include that inside the AI profit boardroom link in the comments and description. Prices going up this month. This is a community all focused on saving time and making more money with AI. So, if you want to learn how to do that, feel free to get this stuff. And inside the SAP section here, we've got tons of training courses on, for example, like how to build anything with Gemini 2. 5 Pro, including how to set up like AR Studio and how to use and build anything with Gemini 2. 5 Pro in 9 or 10 different ways, like you can see. And also we've got a full course on like Deepseek 3. 1, NA10, MCP agents, whatever you want to learn pretty much when it comes to AI is all inside here. So feel free to get that link in the comments description and inside the community you can ask 578 members, anything that you want about AI. So if you get strugg if you're struggling about something or you need a bit of support, etc. And feel free to post inside the classroom and we can help you out. Like you can see right here, everyone gets an answer, right? So if you have a question, post inside the community and we'll get back to you as soon as we can. And everyone's here to just help each other out. Like you can see. So for example, Constantine was asking a question and six people actually just came back and answered. Yeah. And also there are live Q& A calls here. So you can jump on the weekly live Q& A calls, ask any questions you have and we're happy to help you. Top of that, if you want to get a free onetoone SEO strategy session, feel free to get that link inside the comments description. We'll show you how we take websites from zero to 145,000 visits a month and generate hundreds of thousands of dollars in sales on autopilot on this free SEO domination plan. You'll get a custom tailored game plan so you can generate more lead sales and profits from your website. You'll be able to ask any questions you have live on the call

and you'll learn how to 10x your SEO traffic. So if you want more free traffic, more customers, and more sales from your website, feel free to get that free SEO strategy session link in the comments description. Appreciate you watching and thanks. Bye-bye. I'm going to show you how to build MCP servers using Google Gemini 2. 5 Pro. So, the benefit of using Gemini AR Studio is that you can actually build MCP servers or you can code whatever you want. For example, apps, games, tools, etc. Using Gemini 2. 5 Pro, which is the latest update, super powerful. It seems to be on the same level as Claw 3. 7 Sonet, but it's free. And if you want to use this, then the way that I would recommend setting this up if you want to set up like MCP servers and automate whatever you can. Then you can download Visual Studio Code. So we're going to do that in a sec. And then we're going to get an API key from AI studio. google. com. Right? So you just click get API key, create API key, and then from here we can grab an API key that we can plug directly into Visual Studio Code. Right? And there's two ways to set this up, right? For example, you can use R code and that enables MCP server creation. So, for example, you can see I've got Plexi set up and ready to go right here. The other option and I think this may be even easier is if you go to client, client have made this ridiculously easy, right? So, if you go to client, which is an extension inside Visual Studio Code, again, Visual Studio Code is free, client is free, and the API key from Gemini 2. 5 Pro is free as well. You can go in here and then you can install client and go to MCP servers. All right. So, I'll show you how to set up client. You just type in client in the extensions tab over here. Wait for that to load. Then you can download it. Once you've downloaded it, then you can go over to client on the icon over here. Click that and then click settings. All right. Settings over here. Once you've done that, then you can navigate to the MCP icon over here. and they've literally got a marketplace where you can set all of this stuff up. Now, let me show you how much easier that is than using Cursor. So, if we go over to Cursor over here, this is a lot more like hard work. So, if you want to set up MCP servers inside Cursor, you go to your settings and then you go to MCP and then you add a new global MCP server. You have to create the file for the MCPS and then insert the information right there. It's a lot more technical. It's much harder to do. Whereas if you want to set this up with Google and Gemini 2. 5 Pro, you can do that here. Now you might be wondering, okay, why would you use an MCP server? What what's the benefit of this? This is basically the future of AI agents, right? So previously, if you were building an AI agent, it was either not connected to the internet, right? Or if you wanted to connect it to the internet, so for example, something like Plexi, you're only going to get one feature, right? And then you have to set up different tools, different apps, etc. It's a bit of a nightmare. You can use something like for example NA10 to do all this. But again, it's a lot of work. You have to set up this and this, right? And these are all going to be different features of the same tool. So for example, you might have one for sending emails, drafting emails, opening emails, and another one for writing emails, right? It's a lot of work. Whereas, for example, with MCP servers, you basically, for example, you can have an AI agent that's linked to a Gmail MCP and then from there it gets all the functionality directly. You don't have to set up individually. So, it's a lot more efficient and that's basically the future of MCP servers, right? You can pull in loads of good data. So, we're going to go over here and for example, if we want to set up Plexi, we can just type in Plexi Research over here. We can click install. And what you can see is this is going to use an API request to set this up. Now the API that I have set up right here is open router, but I also have Gemini set up as well. But either way, you can use Google Gemini 2. 5 Pro. If you want to get an open router API key, I'll show you that in a sec. So what you would do is you would go over to Openroo. ai. Then in the top left here, you can go to keys and from the key section, you can just create a key and set that up inside the settings of client. Right. So, we're going to retry this. Let's see if it works. If it doesn't, we'll grab a Okay. All right. So, what we can do, like you can see right here, it's a little bit temperamental on open router, but we can grab an API key from Google. Let me do that. We'll copy this one. I'll delete that in a sec as well. Go out to Visual Studio Code. We'll plug in the API key. We'll select Gemini 2. 5 Pro Experimental. We've got some custom instructions here as well. They're inside the AI profit board. If you want to get the custom instructions from me, we're going to retry that request right now. And let's see how we get on, peeps. All right. This should be pretty simple and easy to do. The thing that it's going to ask me for in a sec is an API key from

Plexi. All right. So, if you want to set that up, you can go to your settings inside here. And then from here, you're going to go to API. And then from here, you would scroll down and go to API keys. Generate that bad boy. Copy it. Go over to Visual Studio Code. You can see it's now said, please provide your API key. Happy to do that, mate. There you go. And the other powerful thing about this is like you can give Visual Studio Code or client or whatever you want a connection to the outside world, right? Which means, for example, it's connected to Plexi and it can talk to their server and it can ask different queries and that sort of thing, right? So, we're going to wait for that to load right here. If you want to see what's set up, we'll come on to that in a second. Actually, I'll just wait for this set up. I don't want to mess around here, mess anything up. So, we're going to approve that. Bear in mind, like sometimes it doesn't work. So you can see here, for example, it didn't work perfectly. We're going to have to go back and forth and be a little bit patient with it. Bear in mind, MCP servers are relatively new, right? This is a relatively new feature that just came out a few months ago. And now, for example, cursor Claude were the first ones who originated with it. Rue code, you can set it up in cursor as well. And then also, you've got chat GBT that are planning to launch it directly inside the desktop app. So, wait for that to load right here. see what happens next. So that is running it magic right here. So okay, it exists in the build directory blah blah. Just going to wait for that to load. We'll just hit proceed while it's running. Going to X off that. You can see it's setting something up here as well inside the terminal and then it's going to test it out. Right now we can actually click auto approve and then every time we want to use this MTP from Plexi, it will approve. Again, wouldn't expect this to work perfectly first time round. So you can see it's got another error. I knew that was going to happen before I clicked on it just because of how temperamental this can be. But wait for this to load. And you can see how it's editing the client MCP servers here. Right. So it's got the code for the MCP section inside this document. Now in the meantime, whilst that's setting up, we'll just approve that. let me show you what you can do with this. So if we go to Plexi, for example, you can see all the features. So if you actually want to see okay what can I do with each MCP you can just go to the MCP of that particular place right usually hosted on GitHub. So for example MCC MCP server setup you can see what you can do with it. So you can for example search get documentation find API keys check deprecated code etc. And if you have any problems you just check that. The other cool thing as well if we type in MCP GitHub claude we can see all the MCP servers that are available right so these are all the servers that we have so for example like Google Drive Google Maps you can connect this to Slack as well we're just going to press Ctrl C on that this is not working at all right let's keep going now going open up a new window here. Let's see if it works better with it. Keeps hitting the limit there. That is no bueno. All right, let's try open router. There we go. Going to just delete the old MCP server that we have and we'll try installing Plexi Research from Ruko directly from client. Sorry. There we go. All right, it's working. All right, so here's what I had to do. if you've already installed this. I think this is because I do so many installs of it for YouTube tutorials. But that was messy. That was very messy. All right. If you want the clean way to set up, basically go to R code, open up a new folder, right? So, if you were inside a previous folder with the instructions, then just make sure you open up a new one like I did right here. Then you're going to say set up the MCP server blah blah. You can see here we've got 2. 5 proximal setup. So that's free to use. There's no charge on the API key. And then basically from the marketplace if for example you want to install Plexi just click on Plexi Research, hit install and then say okay I can help set up the MCP server. Here you go. Did the MCP requests gave it the API key like you can see right here. And it's like thanks here you go. Test it out. it were first time, right? So, this if you've got a previous install of this or you've set it up before and then deleted the API key, etc. I would recommend just setting up a whole new folder and just getting started again cuz that was very messy. All right, but the easy way is as I've shown you, just set up a new folder, use Gemini Pro Experimental, and that was using the Google Gemini directly. Okay, took like

minimal back and forth. So, let's test it out. Now, we're going to say, okay, what happened in AI news today? And you see the query here. So this is what it's going to ask to search query on Plexi. And then it's got it magic right there. It seems to be using the think mode as well inside perlexi. So you can see it's okay. Let's fill this out. Blah blah. Here's how we're going to do it. I wonder API key using. But then you got the nicely formatted section right here. Right. So it says, okay, I use PLC's MCP servers search tool to find the latest AI newses requested. This is from today, so March the 30th. And then you got Apple strategy and Samsung right here. But yeah, like I said, learn from my mistakes. All right, so don't do the complicated way. Do the easiest way, which is with Klein. Open up a new folder. So you just go to file, open folder over here, set up a new one if you've got a previous install and then from here go to Plexi, install it, grab your API key, and that is so simple and easy once you set up the new version. So also it's got some interesting stuff here. So we've got, for example, like browser use, AIdriven browser use automation. That's quite interesting. Let's see what this is. Browser use MCP server AIdriven blah blah. And then you can install it on claw desktop as well. Quite interesting local. And then you can see the API keys here you can use. So you can actually use Gemini and Google API key for that. Let's test this out. I wonder if we can get it working. So we're going to go to install browser use. It's probably going to ask me for a Gemini API key. We'll test that in a sec. Hopefully it doesn't break anything as well locally. Looks interesting though cuz the way that I know to use browser use with client directly is you have to get a clawed API key, right? Which is not cheap for most people watching this. Like it can cost a bit of money, but it's setting that up. So, let's see what we get back here. MCP servers updated. Updating MCP servers. So, it seems to be working. And then what we need to do is just grab an API key. It's hit the limit as well. So, go back here. Grab an API key. Copy that. Have to delete all my API keys again in a sec. All right. So, that's good to go. Let's approve that. So, it's trying to open it up, but it can't. Let's see what we need to do here. We might need to give it a new API key directly in the chat here. For now, what I'm going to do just to make this simple, I'm going to switch to anthropic. Use the API key from Claude. Retry that. And then for the actual browser use, I'll use Gemini. So you can see it's progress here. If we open that up. So we'll start a new task now because it seems to hit the context limit. Yeah, that is wouldn't recommend using the browser use MCP does not work. But Blexis does though. You see that's working beautifully right there. All right, let's try some others. So you also got for example send grid which you could probably send emails with. What else we got? Fire cruel advanced web scraping and crawling wolf ram alpha air table could be quite useful. Ser search as well. I think that'll be really good. So you can basically use this with Googlelexi. The good thing about flexi is like you get the AI results not just a Google search which is pretty good as well. But yeah, you can see how to set them up. So simplest and easiest way seems to be you open up a new folder then you install it from marketplace. Bear in mind like not all of these seem to work. So for example, browser use that says installed but it totally let us down right there, right? Whereas the Plexi one much better much better. You also got for example like Google Calendar, Discord, Spotify as well which is pretty useful. I wonder if that requires a API key. Let's have a look here. So you can read more about these if you want. So, for example, if you want the Spotify API credentials to link MCP servers to client, then you would just go to the developer dashboard and start getting that set up, right? Like you can see right here. And then installation looks fairly simple, but you never know. All right, but the most reliable one that I've seen so far is Plexi. And then also Brave Search, I've tested. That's pretty good. Airbnb I've tested inside NA10. I've got other video tutorials on that. But that's how you can set up MCP servers. And then you can see here you get the results back and you can set all this up with Gemini 2. 5 like I was showing you. Now if you want to get my community that shows you how to make money and save time with AI, feel free to get that link in the comments description comes with lots of tutorials on MCP agents. Right? So if you scroll to the left here, you'll see our workflows and tutorial on MCP agents right there. And also NA10 MCP agents

which is a lot more complicated than just using the straightforward client method where you can install directly. You can also go in the calendar and jump on the Q& A, ask any questions you have. And also inside the community here, if you get stuck in between different Q& A calls, you can post inside the community, ask any questions you have, and we are happy to help you. All right. And if you want to get a free one-to-one SEO strategy session that shows you how you can make more money and save time and get more customers with SEO, feel free to get that. We'll show you how we take websites from zero to 145,000 business month and generate hundreds of thousands of dollars in sales and optimizers. Feel free to get that link in the comments description and thanks so much for watching. Today we are going to be talking about the latest update from Rode and how to use it for free with now this is a completely different way of programming AI agents between coding architecting debugging it also help you save credits and I'm going to show you exactly how to do all of this for free step by step so you can build apps tools games websites etc using this new feature now I've never seen anything like this come out from any other sort of coding program for example like Wind surf and other tools like that. Don't think you can do this. Now, this is a brand new feature and essentially what it does is it's going to down any sort of task whether that's building or coding an app, a tool, a website, etc. throughout the the sort of process. It's going to break it down into separate tasks. All right. So, let me show you an example. We got the legend himself, Vic, on the call today. Welcome, Vic. — Welcome. Welcome to me as well. It's great to be here, Julian. Super excited. You know, this is something super valuable for the people in the AI profit boardroom. They've been asking stuff about R code. So, I think this is going to be an awesome addition. You know, segmenting tasks and all that. Super fun. — Yeah, it's a game changer. It's basically like you have a swarm of army, you know, like an army of AI agents that just go off and like tackle really complex stuff and break it down into like little multi-step projects. So, it's going to be a game changer. And I know a lot of people in AI profit bordering were asking about this, too. So let's get through it and I'll show you exactly first of all how it works and then I'll show you how to set up. All right. So for example, if we go inside Visual Studio Code and then we go to Rode, which you can see on the left hand side over here. And by the way, if you don't know how to use Visual Studio Code, you just go to code. visisualstudio. com and then from there you're going to go over to extensions, select Root Code. All of this is free, right? But it's a super powerful coder. I know you've tested, for example, Vic, some like browser extensions like it can go off and use your laptop and that sort of thing, right? — Absolutely. You can even book a flight for myself, you know, to the Maldes or something. So sometimes I'm scared it go it puts me on a vacation that I'm not looking forward to, but you know, it's pretty good. I'm kidding. It's awesome. — I see that you could end up in Zimbabwe or anywhere, right? — Absolutely. — All right, cool. So you can see an example here. So basically what we said inside this task is like create a sleek conversion focused landing page for Goldie agency's AI automation dashboard. The design should be modern fast loading and highlight the benefits of AI powered business automation. Right? So it's a pretty complex sort of strategy right there. Um essentially what you can see here is it's going to flick for you right you don't have to think about whether you want to use code or architect or ask or debug which I think nobody can be bothered to switch between anyway. and it will give you a full task like this and be like, "Right, first we're gonna do this, right? Then we're gonna do that. " And it just breaks it down really nicely into it. So, it's pretty powerful. And then what it's going to do is it's going to create like separate subtasks, right? So, you can see here how we gave it the prompt and then 2 minutes later it's like, "Right, I'm going to launch the landing page for you here. " Um, so it's is pretty cool because it's like it's creating a shorter context window and breaking down each of the tasks into subtasks. So, I kind of show you an example of what you'd normally do. Like, normally you're going to brute f brute force a task and you all in the same context window and sometimes you can use like thousands of tokens, right? Um, and it gets quite expensive and that sort of thing. Whereas, for example, with this, you give it a task as a prompt, then it's going to break it down into loads of little subtasks that and the agent carefully figures out, okay, which subtasks for this one, that one, etc. And eventually between all those you get your output, right? So you just save tokens and you take a really complex project and break it down into little tasks. And this is just with one single prompt if that makes sense. — So you're basically saving a bunch of money as well, right? By saving those tokens. — Yeah, 100%. Or if you're on the free API key, so if you're using Gemini, which is free to program with and code with, but runs out of credits very quickly, this is like a more efficient way of using it, if that makes sense. Absolutely. Yeah. — Yeah. So, let me show you how to do this. Basically, what you do is like, okay, let's say for example, we want to create a website, right? We can say, okay, create a website for my SEO

agency inside here. And we're using Boomerang Rotes here, right? So, if we hit enter like so, and what I'm going to do is just make sure I've updated my API key. So, I think I deleted one on the live stream the other day. So, I'm just going to go to But this is really good for anyone to see, but I'm going to go to a studio. com. You can get a free API key from Google using the latest knowledge and that sort of thing. Copy that over. Plug it over to Visual Studio Code. Plug it in settings over here. Hit save. Hit done. Boom. All right. And then it's going to attempt that in like 14 seconds. So I'm just going to redo that. We'll hit resume task. And there we go. Boom. All right. So it starts using that. We're just going to hit allow over here. And then we can start using that process. All right. So I'm going to like it really didn't actually break that down into like look on the settings over here and then we'll say create big website for my HD with all my etc. Hopefully that should start using the boomerang mode. I'll show you how to set up the boomerang mode in a minute. Vic, if you get if you see any comments on the live stream etc. Do you want to just read them out as we go? So we make sure we answer everyone's questions. — Absolutely Julian. I'm paying close attention to it so don't worry. — Beautiful. All right. So here's an example. Right. So we gave it the prompt. Let me X off this. So we gave it the prompt like give create a big website for SEO agency with all my services etc. And you see here how it's like okay creating this is a significant project. So let's break this down. Um and it switches into ass mode. And then what it's going to do is it's going to use that boomerang feature to turn everything into like little subtasks. So you see it's like right it's kind of like a consultative approach where it's like right how should we define the requirements for your SU agency? Do you want to gather some information, a list of pages? Do you want to start with basic structure and ask details later or just cancel the task? So, go ahead. — We do have a question from Rod Julian, which isn't really tied into a root code, but the man asked, "When do you sleep? " And I think that's a genuine question that I'm also interested in. You know, — I actually slept at 8:00 p. m. last night, and I woke up at 6:00 a. m. I was exhausted. But usually, yeah, usually I sleep like 8 hours a day. So, I sleep super early. I don't drink or smoke or do anything else which just allows me to like sleep super early and get stuff done, you know. — So, basically, you are a machine, right? You — That's it. I'm literally I'm a machine. I wanted to be modest. humble. But, uh so yeah. No, that's it. Prioritize sleep, people. You got to stay healthy, you know. — So, I agree. Sorry. — And AI should allow you more time to sleep. We should do a sleep course inside the AI profit board. — We should. That would be good. — All right. Cool. So, we're going to kick it off now. And I'm going to say all right so we can for example provide like a list of pages and services or we can start with a basic structure. So just to keep things simple I'm going to start with a basic structure and then add the details later because this you know it's for a YouTube video. So you see how it's switching now into code mode and it's created a new task. Just created a new task out of nowhere you know. So that's a new task. Like you see our prompt has disappeared completely. So if we go back to this like it's already created like one task right there and we already gave it a prompt. So we've got two different tasks going on. So if we go back to Visual Studio Code over here and then we open up a new window just so you can see what this looks like. We'll go back to R code. What it would do is create everything inside one long prompt. But here is like task one, task two, etc. And we use less tokens doing that which is pretty nice. So if we go back here, and by the way, we're doing all this for free. So we're building out like full websites for free, which is pretty nice. And you can also combine this with MCP agents, which we've got a full tutorial inside the AR profit boardroom. That's essentially how this goes. It's awesome that you mentioned NPC agents. We actually do have a question from Amazon Andy. He said, well, it's like a three-part, so I'm just going to read them all out for you. The first thing is, how does this compare against Manis, Julia? — Yeah. So, I mean, for me, the only Okay, so I mean, number one, this is free, right? Whereas, for example, with Manis, you're paying $200 a month, right? Number two, it's becoming more like Manis because with Manis, what it would do is like create loads of little subtasks inside one chat. And that's essentially what this is doing, right? Whereas previously what Manis would what you would do with client is like have one really long chat, which didn't really make sense. So, let me log into Manis. I'll show you what I mean. We're just going to log in here and sign in if I can. I might not be able to get in here. Let me try and sign in. We'll try and sign in with that one. They might have kicked me out. — Imagine. No way. — It could happen. It could be possible. Seems like I can't log in, but we'll come back to that later. So, if we go back to Visual Studio Code essentially. Yes. So, there's no — you don't need to pay for this because you're using the free Visual Studio Code. The extension is free. Root code is free and also Gemini is free too, right? So, that's a big difference. And then I would say the main benefit of Manis is like it can deploy full-blown apps uh a subdomain if that makes sense. So like — you can say to manus okay create a website from SEO agency and then deploy

it right and it will deploy it on a manus. space subdomain which means that you know you get that response back whereas root code you can't do that you can't deploy to a subspace — a subdomain. — Yeah. And then additionally a lot of the other thing is as well like inside ruode you can actually set up a debugging version of chrome and then chrome can be used with root code. So you can get claw 3. 7 sonet to start using your browser if that makes sense. — Absolutely it does. — Yeah. Whereas for example with manis it's like a separate terminal. It's kind of like a completely separate environment which means that it's very hard for you to log into things because you have to give Manis all your login details or take control etc. you know. — Yeah. A lot of credentials. — Yeah. That's it. So and the other thing as well like you need a Manis invite right to be able to use Manis you need invite like it's exclusive. So not everyone has access whereas you can download root code today and then in your settings here if you go down to the browser use section here you can get it to take control of your laptop. So you can set up a local IP run Chrome debugging and it if you're listening to this you never used that before like it's going to sound super technical. So just check out the tutorial inside the AI profit bordering but essentially that's how you do it you know. — Absolutely awesome. So you can see here now what it's doing is like it's like what should we work on next and so you can say okay improve the overall visual design add more details refine the content add a new page etc. It just breaks everything down step by step for you creates like a really customized workflow. The other thing is as well like previously I'll try and find an example but previously what would happen inside client is like it would get stuck in a loop because it was running off a big context whereas this is creating separate tasks. Well, for example, it's creating a let me view some previous ones where it's broken before. H I think it'll be hard for me to find. But essentially what happens quite often inside recode is like it can get stuck in a loop because it's focusing on the same contact. Whereas with this like it just goes off and does its magic in separate subtasks. So it's less likely to break and less likely to go into an error loop. So it's pretty cool update. And then also what you can do is like set up MCPs inside here and then you can have your AI for example do one subtask where it searches perplexity or um Netlefi and stuff like that you know. — Yeah. So it's pretty cool. And then the other things I mean the other things that you can do inside root code as well is like you can actually set up claude as the API to start using your browser or previewing the website inside the root code as well. So there's a lot of stuff going on here. You can see it's hit the limit. So if you ever hit the limit inside Google Gemini, what you can do is you can switch between models here. So you could say, okay, just use flash instead now. So we've hit the limit on Google Gemini. And you can see it continues now. Or the other thing you can do is you can go to your settings and switch between open router and open router will give you more credits there as well. So if you hit the credit limit, you can do that. So I'll set you I'll show you how to set up Boomerang. Okay. — Yeah. Before we jump to Boomerang real quick, uh I saw a couple of questions actually because you did mention MCPs and people are you know all about the buzz. Do you have like a favorite uh MCP use case or which MCP you think is the most like let's say the best one or even top three if you have you know we even just had a donation asking for specifically favorite MCP so thanks so much for the donation — absolute legend thank you yeah so I mean my favorite personally perplexi is really good simply because with perplexity you know you've got you got Google Gemini you got Claude but they don't even know what's going on today right and the data is outdated, right? The biggest limitation of any AI, especially if for example, you know, you want to create content around what's happened today or you want to create a new website around what's happened recently. You know, information gets outdated really quickly. So just linking this to perplexity allows the AI agent to not just build a website, but also to Google, you know, what's happening in real time, what's happening today. It can go off and do deep research reports as well, if that makes sense. — Absolutely. — Yeah. Let me show you some of my favorite. We'll go into client here and we'll open this up. So I'm just going to close Visual Studio Code and then we'll reopen this up. Right. So if we go inside client here, I'll show you some of my favorite MCPS. So client is probably the easiest and best way to set up MCPS, but you can see all the stuff that you can set up here. So I would say Plexi is my favorite. Just type in Perplexi and you can install that pretty easily right there. Um, what I also really like is Brave Search. Again, for the same reason that I do a lot of SEO. You know, we SEO for our agency as well. So, this is great for just doing SEO. And, you know, you can even create like a content outline tool or content creation tool that searches the web, uses Brave Searches API, which

is free, and then uses that to like reverse engineer what's ranking, what content people have used, and then how can we make a better way or a better piece of content based on that, you know? And the other thing as well like you know between brave search and also between perplexity like if you go directly into claude and you're not in the US then you can't use the web search feature right and so claude is very limited if you're like okay write some about wind surf 6 for example right windsurf 6 just came out it doesn't know what it doesn't even know what windsurf 6 is right it thinks that it's about water sports right whereas for example If you can connect it to perplexity and we go into perplexity and we say okay write about wind surf 6 because perplexity is connected to the internet it can actually understand okay wind surf 6 is about AI here's the latest AI updates and this is what it's about so you just get much more accurate data that's less likely to be hallucinated and you make your AI agents way more powerful inside uh root code or client yeah I think Shopify if you're running the e-commerce store Shopify is by far the MC MCP you can set up because you can integrate your products with whatever you build inside client and the final one that I would say is uh yeah so we we've talked about perplexity brave search is really good for e-commerce owners and then I would say air table as well air data and that sort of thing you know — yeah cool yeah so it depends what you want but it's super easy to install and set up and that sort of thing and then you can connect the power of root code and it's Boomerat tasks with MCPS — awesome — awesome All right, cool. Yeah. So, if you want to set this up, super simple. So, what you're going to do is you are going to go, let me show you here. We're going to go back to the guide. All right. And essentially, you can download this JSON file and insert that into R code. So, I set it up this morning. It was super simple and easy that way. Just trying to find exactly where it is. So yeah, inside you just click on root code here. Then you're going to click on the bottom left, hit edit, and then inside here, you can insert the custom rolls, right? So you see how we've got boomerang selected. You put the roll definition, you grab that from the boomerang guide here. So you copy this and then you go back into R code, plug that in, and then inside the model, inside the mode specific custom instructions, you're then going to grab this. So the custom instructions from here, plug that in and then paste that in like so. Then hit done. And you can switch between boomerang and all the other features here. Right? If you just have boomerang set up uh by default, then it will switch between code architect ask and debug by its own. Right? So you can figure out everything by it own. — There is a question Julian. What do you think would be the best use case for let's say boomerang? like do you have any off of the top of your dome that you think oh this is a game changer specifically for this? Yeah, I think anything where it's really complex and you've got a lot of subtasks, right? So, if you're for example like trying to create a big website, then it's a game changer for that. Anything that requires a lot of research to build as well, super useful. But basically, if it's super linear, like if it's just build me a SEO calculator tool and it's just like from point A to point B, you don't need this. But if for example you're building out a big website with lots of pages, categories, if for example you're building out a very complex app where you need to break it down into little subcategories, this is what you want to use. — Makes complete sense. — Yeah. Let me show you an example, right? So you know you can use something like data button. Data button is a super powerful tool for doing stuff like this because it will flick between like architect and all those sort of things. So let's load this up and I'll show you what I mean. Like if you were building a job board for example. — Mhm. Then you could check that out. Let me sign out here. Yeah. So just a super complex project. So I'm just trying to make sure I log into the right account on data button. One sec. It looks like all my apps have gone. But essentially, yeah, if you if you're building anything big, then this is perfect, right? So it's just orchestrating like really complex tasks and workflows, if that makes sense. — Yeah. Yeah. — Uh, we do have Val donating again asking, "Have you tried prompt by DD devs? " I think he means like a MCP agent prompt. — No, I've not tried that, but I would like to check it out. Maybe we can do that on another live stream. That would be cool. — Um, one t one tip I will give people today is like if there's one tool you check out today, this one impressed me massively was Lindy. Lindy was really cool. essentially on Lindy you know like we we've set up Zapia tasks together right? — Yep. — And you know it's like super complex and you have to plug in all your APIs and all that sort of stuff. Basically inside

here all the APIs are plugged in already and all you need to do is just connect your email address and then you can automate for example like super complex workflows like this. You just hit click on add your Google Drive sheets and stuff like that and it's pretty cool. Like for example, this morning I built this tool and you'll see the video on my main YouTube channel. This tool right here, it basically automates like customer service. It was super simple and easy. Um, you can add custom knowledge. So I added our website to the custom knowledge base here. — Mhm. — So you can add websites or text or files, whatever. And then you know you you've created an automated chatbot. Customize your website in like a few clicks. Super simple and easy. Yeah. Um, so like this morning with Lindy in just a few clicks, we automated, let me show you inside the AI profit boardroom and we got the templates inside there. So we automated email drafts and on the website like without any coding and you know like NA10 can be really confusing to use. This is like a tech this is a much simpler way of using that. — Awesome tool. — Awesome. Yeah. So check that out. I've got a video on the main channel if you want to check that out and learn about it. But yeah, any other questions before we go? I think that's about it. A lot of people dropped some uh like interesting their favorite, you know, MCPS such as desktop commander, fetch, web search, etc. But yeah, overall just yeah, very good and awesome tool when it comes to root code and its boomerang feature, you know. — Nice. Yeah. So, I mean like just to recap here in case people are not 100% clear on like why would you use this? Well, number one, it makes like complex tasks basically put into little subtasks, you know, like you would do on Trello or Asa and that sort of thing. So the AI figures out exactly from your main prompt what needs to be delegated and which tasks need to be broken down. And then also it will automatically figure out the right mode for that. So whether you need to use for example like architect or code or ask or debug depending on what the task is. The final thing is like it just keep it keeps the AI super focused on like that one subtask instead of doing loads of different ones and it makes the results like way faster and smoother. Awesome. All right, cool. So, if you want all the templates and all the workflows that we use at our agency, Vic, I mean, how many have we set up over the last 30 days? It's insane, right? — I'm not going to lie to you. I think we have over like hundreds of lessons by now and it's growing exponentially day by day. So, it's pretty awesome. — Yeah, it's insane. And if you want to ask questions, jump on the weekly Q& A. Vic, you did one yesterday and absolutely smashed it, right? Like I saw all the feedback, people saying like how good it was and everything. — Yeah, we had over 20 people join us yesterday. It was an amazing vibrant Q& A. We even had to go one minute overtime, which usually, you know, in Q& A it doesn't happen, but it was so good. A lot of great questions. People were interested in the most basic of N8N setups, and I showed them like from the most basic of N8N setup, how you can connect it, what you do with the nodes. It was pretty awesome. So, if you get in there, every Wednesday we have a Q& A. Feel free to join and I would love to welcome you. — Awesome. And then if you get stuck in between the Q& A, just post in the community. You can see here, for example, Vic, you're always posting new updates. It's insane how many updates you post. And then like I mean this, for example, Denise just posted like two days ago. You can see all the answers and the help they got. Like it's just a community full of great people helping each other out and I respect that massively. — Absolutely. — Awesome. Yeah. So, feel free to sign up. I think after this video the price is going up, right? So, if you want to sign up, sign up right now before the prices go up cuz literally we'll give everyone like 30 minutes and then we'll put the price up again. But yeah, it's just an a community focused on making money and saving time with AI. And then additionally, if you want to get a free one-to-one SEO strategy session that shows you how we get thousands of visitors to our clients websites and if you want to customers or maybe you're tired of spending money on ads, maybe you want exponentially, then feel free to book that in. It's completely free link in the comments description to get a free SEO strategy session. Anything else you want to say before you go, Big? — No. Look forward to meeting all of you in the air profit boardroom. This was awesome, Julian. I learned a lot about, you know, Boomerang and Root Code. Can't wait to test it myself. But yeah, perfect. Perfect execution here. — All right. Appreciate it, man. Cheers. — Gemini 2. 5 Pro versus Chat GPT3. Who is the winner? Today we're going to be testing out a bunch of prompts and seeing which one gives the best outputs. Now 03 is just been released this week. Gemini came out last month. A lot of people already saying that Gemini is the goat of coding. We'll test that out today and just see, okay, who's winning, who's the best, which one gives us the best outputs in real time live. So let's get straight into this. Now, if you're wondering, okay, how do the benchmarks compare? For example, we'll run through these in a second. This is an amazing website by the way, docsbot. ai. And you can compare both models. Obviously, Gemini 2. 5 Pro comes from Google. And you've got 03 that comes directly from Open AI. Now, if we compare the stats right here side by

side, you can see for the input context, window 03 allows 200k tokens, which is significantly less than Gemini 2. 5's Pro tokens, which apparently is also going to be introducing 2 million soon, which is pretty insane when you think about it. Terms of maximum output tokens, 64K for Gemini 2. 5 versus 100K open source. Obviously, both models are not open source. And then the release date when the model was first released. So, this just came out a few days ago on April the 16th. And on March the 25th, 25 days ago, Google Gemini came out. The knowledge cutoff date is quite significant too. So, you can see here 03 cut off May the 31st whereas Gemini's cut off date is January 2025. However, you can for example inside chat GPT just enable search and then it can search in real time based on what just happened. All right, so you can connect 03 to the internet using that method. In terms of API providers, you can get these on open AI API but you can also get them on open router as well. We'll be testing like how it performs in terms of the API but also directly inside the chat here. And then pricing comparison, this is a huge difference. So, in terms of input tokens, right, the cost per million tokens from 03 is $10 per million tokens, which is way more expensive, almost 10 times more expensive than Gemini 2. 5 Pro. And the output is $40 per million tokens versus $10 per million tokens from 2. 5 Pro. Big difference in the token costs here. You can also see the benchmark comparison. Maybe we'll show that in a second, but I don't want to bore you to death with the statistics. So, let's get straight into testing out which one performs best in terms of creating stuff. So, we're going to be testing stuff like games, science, onepage websites, SEO prompts, tools, and full websites as well today and just breaking down these bad boys and seeing which one performs the best. By the way, all these prompts, tips, workflows, etc. inside the AI Profit Boardroom, the one and only link in the comments description. So, let's get straight into this. What we're going to do is we're going to plug in this basic prompt which is just make me a captivating endless runner game. Key instructions on the screen blah blah. I like pixelated dinosaurs and interesting backgrounds. We're going to get straight into something else which is creating a self-playing soccer game. All right, so let's get straight into that. We do the same inside both inside 03 as well and we can see how they perform. Now, one of the first things you're probably going to notice here is that from what I've seen, let's I don't want to say anything too early, actually. I'm just going to let this see how it rolls out. We'll see how it performs. But from what I've seen so far, 03 seems to be faster, but we'll see how they perform. All right. So, we're getting the p5 js code back from both of them right here. So, let's see what we get. And then we'll run that through the p5 js web editor and we'll compare them side by side in a sec as well. Right, here we go. So, we got the code back from Gemini Pro Advanced. Let's copy that and we'll test out inside P5GS. Hit play. And we have a playing soccer game. It reminds me of Football Manager back in the day, but it seems to be repeating the same loop over and over again. Let's see what happens next. Ah, the Blues are absolutely destroying them, but it's a bit weird. If you look at the way it's simulated, it's like they all just gather together in a weird block, which is quite strange. Now let's compare the output from 03. All right. So this is the output from Gemini and then let's grab the output from Chachi Petite 03. There's not even a goal. Look at that. There's no net for this is really weird. So if you look at the Let's stop that and I'll play it again. You see how there's no net on this side of for the football match. And also it seems super buggy. Like where's the red team gone? It's only blues now. So, if I had to compare them side by side, I'm gonna say, and this is Chat GPT's output, for sure, Gemini 2. 5 Pro wins by a long way there. And a lot of people saying from their test as well, this is not just my opinion, but a lot of people are saying from 2. 5 Pro, it's the best of tested for coding and analyzing just in general. Flash, if you've not tested it before, so 2. 0 Flash just came out yesterday. If you've not tested that before, then I would definitely recommend checking out my video from yesterday about it. Honestly, I was very underwhelmed. I don't want to hype it up and BS you and tell you it's the best thing since sliced bread and fish and chips if it's not. And it certainly isn't, mate. So, I'm not going to use 2. 5 Flash again. I think it didn't impress me at all. All right. So, let's keep going through now. So, you can see here so far. And I'll make a little list of this if we put this on Google Docs. So, in fact, we'll do this on a cheeky Google sheet. So, it just looks a bit better. All right. If we had to compare

him, who's the winner for the football test, the football simulator game? By far, it is Gemini 2. 5 Pro. All right, so we've got the winner for the first test. Look at this beautiful spreadsheet. This is years of spreadsheets paying off right here. Let's delete that. Pop those in. So, test number Uno. We have got the football simulator game Gemini 2. 5 Pro won by a long way. Right there you can see it with the UI. Let's keep going now. So we're going to test out creating a 3D car simulator. All right. So let's plug these in. We'll start a new chat just to get it rolling and remove the content. Let's see what we get. Now the one thing that I would say is a caveat to this is like we're using these prompts, right? But depending on which AI you use the most, there is a clear winner for self coaching. Right? So if you need to help making business decisions, if you ever coach yourself, like for me, I do a lot of reflecting and journaling and that sort of thing, Chad GPT is amazing, particularly 03 for getting like really good advice based on what it knows about you, right? So for example, if I go into chat GPT 03 and I'm like knowing what you know about me, right? because it remembers pretty much every conversation we've had. Give me some advice or tell me something I don't know or for example an help me answer this question. What do I need to eliminate to move further on my goals, etc. 03 is almost AGI level for that. I would say it's absolutely amazing for that. I'm not going to show it cuz it's super personal and it's going to be super personalized to me. But I will say that for self coaching 03 wins hands down. But maybe we'll pull that up in a second, but I'll try and keep it like private and that sort of thing. So let's copy that. And this is from 03 for the second test. And the second test is the car game. All right, car game. So let's see what we get back. We're going to run the output from chat GBT3 first. So let's see what we got. And that is doing it inside HTML. So I'll run it through LiveWeave, not P5GS. Here we go. And we have a blue screen. Honestly, I think I tested this a couple of days ago and it gave us a way better output. But from what I can see, that is terrible. Absolutely terrible. All right, that is giving us nothing. All right, so let's go back to Gemini 2. 5 Pro. I want to see is this giving us Maybe this is in P5 JS. I just need to test this out. No, it's not working in P5 Jets. All right, so let's go back to Gemini. Gemini says it can't help us assist on that, but it's Why does it stop there? It's failed the test. I'm going to give Pro another sort of just another opportunity because for some reason it got confused there. And I'll run it through AI Studio instead because it seemed to get confused inside the chat. Let's see what we get back. A lot of people saying it's good that Google is the best. I'm interested to know more on that. There's some people prefer Google over Chat GPT and OpenAI. I'd be interested to know more on that. For me, like chat GPT and OpenAI in general, it does get a lot of hate. A lot of people hate on them. So, I'm interested to know why that is. The other thing is as well, right, if you're using Google products a lot like Google Drive, Gmail, etc., then you can imagine as these models get more advanced, you're going to be more and more in the ecosystem. For example, the iPhone and the Mac and how that all syncs up. It's the same with the Google ecosystem, right? If you run your business on Google, then it's going to be much easier to integrate that. It's like a soft thing. I can't measure any tests, but yeah, I would say for sure like that's going to help and that's only going to advance more and more over the time. All right, so here's what chat GBT gave us. Chachbt03, he gave us this the most advanced OpenAI model in the world. Now, here's what Gemini 2. 5 Pro gave us, right? A million times better. All right, so let's see what we got here. Oh, it's just incredible, isn't it? Look at that. We're flying through. We're roaming around. We're having a great time. It must be a Tesla. Those buttons don't work, but I'll let them off because the game actually works on ChibiTs. All right. And 3D graphics are not bad at all with that little train going around, roaming around. You can't argue with that. How that goes pretty far back as well, doesn't it? You got a nice little circular train route right here. Beautiful. Let's try and find the train. That is amazing to be fair. Look at the landscape. It's just incredible. We're just zooming around and we literally build that in one single prompt. How good could you make that game if you actually spend a bit of time on it. So for sure Gemini 2. 5 Pro is just absolutely steaming ahead right now and 03 is struggling my friends. So let's try something else. Now we've tested the game prompts. I'm not going to do too many more games cuz that's not what makes you money in life. What we're going to do next is we're going to try and create a onepage website. So, I'm

going to say generate a high converting landing page blah blah from SEO agency. I'm going to see if they can actually research information about me as well, but we'll come on to that in a second. So, if I put canvas on right here, we'll put the canvas here and we'll see research Julian Goldie and his agency. Then generate high converted landing page for his SEO agency blah blah. We'll do that and we'll do the same inside chat GPT03 which so far has let us down quite considerably but we'll give it another cheeky. The good thing about this is I can search the web right so with chat GP3 which is probably going to make the outputs better but we'll test that in a sec. We can do grounding with Google search as well. So we can use Google search inside here. So I'll do that inside AI studio. I'll use Google search as well to use the same prompt and see what we get back in a sec. I like the fact as well that chat GBT brings up the canvas. So you can integrate the search with the canvas. I'm pretty let's have a look. I'm pretty sure Google Gemini advanced inside the chat here where you get the canvas option. You can't connect to Google search as far as I'm aware. It's just using it knowledge already. If we go inside AI studio which is using Gemini 2. 5 Pro as well, it should be able to research Google. You can see here we're grounding with Google search and then we'll see what we get back. Some people were saying that Grock would beat Gemini 2. 5 Pro. And to that, I would say check out my video comparing Grock and all the other models versus Gemini 2. 5 Pro. Personally, I wasn't impressed. Like the first day I started using it was amazing. And then after that, it seemed to get dialed down. Like it just didn't give me the same outputs as that. I wasn't that impressed. I just don't use Grock anymore for that reason. Compared to Gemini 2. 5 Pro and Chachi Petite, I just don't think you get the same outputs. So, can we run this? I don't think we can on the canvas, but we'll run this inside LiveWave anyway and see what we get back. So, let's plug that in. Here's the landing page. It's not bad at all. Let's see if it's got any information about me. It's got a little bit of information, but it's still quite generic. Organic growth, authoritative link building, technical SEO, content, blah blah. It's still quite generic information, right, that it's finding you. To be honest, you're going to get much better outputs from something like Gen Sparker Manus if you do this job. But it's not bad. Let's have a look what we got back from chat GPT03. So, let's click on preview. See what we got. Well, like that. I would say both of them were pretty poor there. I'm not going to hide this up and tell you like this is amazing. This is super basic. Look at that. I'm a little bit disappointed. Come on to coding with the API. Maybe that will give us better output since R code in a second. But these are not great. And I think that's actually unless I'm Yeah, I think that's a madeup hallucinated testimonial as well. I don't think that's a real testimonial as far as I'm aware. We have a lot of testimonials, so it's hard to keep track, but that seems totally misleading. If I'm going to compare them, I'm going to say that Gemini 2. 5 Pro created a much better landing page. Like, it's longer, it's better designed, the testimonials are actually real. For example, like this one and this one. This one too, it's using real case studies and the design is much better. Plus, there's more content on the page. Right? If we have a look at the chat GPT output, super basic, doesn't look great. You would never publish that, right? You never in a million years. So, I'm going to go with Gemini 2. 5 Pro again. All right. And you can see here Gemini 2. 5 Pro is absolutely crushing chat on all these benchmarks so far, which is pretty wild. Let's go on to the next task. I've never actually seen a test this unanimous. It's unbelievable. So, let's come on to another one. We're going to say, okay, create a beautifully designed snow day calculator inside both of these. We use Gemini 2. 5 Pro inside Canvas. By the way, someone was asking me yesterday on the live stream like, "How do you get the preview on Gemini? " So, if you're using Gemini Advance, which is a paid version, and then you click Canvas over here, then you can start using the canvas inside 2. 5 Pro. It will give you an actual preview. But it's so far from what I can see, Gemini 2. 5 Pro is just absolutely destroying chat 3, which is blowing my mind cuz this came out nearly a month ago. So, we've got the snow day calculator back from chat 3. It's not bad. It works. Let's test this out. Need to find a US zip code right here. All right, let's take that. It's not working, is it? I'm clicking calculator here, but nothing is working. All right, let's check out Gemini 215 Pro. So, straight away, you can see from the UI, right? Look at this. You see how there's like a nice little icon here for the snow day calculator and there's like estimate the likelihood of snow day. Like the content itself is written much nicer. This calculator actually doesn't work from Chad GPT. So, I've tested clicking calculate. Nothing happens. Now, let's go over to Gemini. The writing is way

more humanized in terms of how it's written. And then we'll put the form in here. Hopefully, it works. Calculate probability actually works. All right. And it's actually, look at the way that it's written. Get the sleds ready. The snow date is looking very likely. If you compare that to this one, like number one, it doesn't work, but number two, like the content is really just AI. It feels like AI. There's no life in it, right? And then if you look at the way the heading is designed, just little touches like the icons here make the Gemini Pro calculator a lot better. So, Snow Day calculator, Gemini 2. 5 Pro is just absolutely destroying Chad GBT. It's absolutely redonkulous. All right. By the way, if you want all these prompts, just get them inside the AI profit bordering. This also comes with a community focus on just making more money and saving time with AI based on how we've implemented all our systems. So a lot of people talk about using AI. They talk about how to scale with it. But there's nothing practical in there. Whereas for example, we've actually taken all the systems that we've used to grow and scale our business. For example, we reach millions of people on YouTube each month. Twitter each month. We have a big email list, etc. And we've automated all this stuff using AI and all the systems right here in the crash course. So if you want to steal our best systems, get them inside the AI profit boardroom along with weekly Q& A calls, all our best SPs and an amazing community of people. You can have 670 members here. Any questions you have about AI and then we're happy to help you. Like you can see inside the community, everything is getting an answer quickly. All right. So, let's jump straight into this and we'll get back in the reins here. So, let's pull this up. What we're going to do next is we're going to try and create like a five-page website. This might take a little bit longer to test out. So, we're going to go inside Visual Studio Code over here. And I'll use Root Code for doing this. All right. And then we're going to go inside the settings and we'll select Gemini 2. 5 Pro to begin with and just test this out. All right, I'll use the paid version just so that we don't run out of tokens, but you can use the free version on open router. That's another thing here is like you can actually code for free with Gemini 2. 5 Pro, whereas you can't use chat GBT for free, right? Unless you were rinsing the trials that I was telling you about last week for Optimus Alpha. So, we're going to say, okay, create a fivepage website for Julian Gold's SEO agency, blah blah. We'll run that through root code here and we'll see what we get back. Now, if you're choosing between using the chat interface or you're using like R code and Visual Studio Code, by the way, I have SOPs on all this stuff below, then what you're going to find is that for creating bigger projects, this is much better to use inside Visual Studio Code. However, if you want to create like just a onepage landing page or a quick calculator, etc., then it'd be much better to use the chat interface directly. It's going to be quicker, it's going to easier, you're going to be messing around less, etc. Now just whilst we're waiting for that to load, you might be wondering, okay, what are like the technical benchmark comparisons like in terms of performance metrics. So if we have a look here, Global MMLU is not available for 03, but for Gemini 2. 5 Pro, it is 89. 8%. You can see for diamond science, Gemini 2. 5 Pro is just outperforming 03. For competition maths and Amy tests, Gemini 2. 5 Pro, humanities last exam 03 is winning. S swe benchmarks, so software engineering capabilities. 03 is outperforming Gemini 2. 5 Pro, but I just I don't see that in my own tests. And then MMU, it's a massive multi-task, multimodal understanding, they're tied, but 03 is beating it by about 1. 2%. But again, I don't really honestly I don't trust these benchmarks. Only trust my own tests. I always think that's the best way. So, let's see what we get back from 2. 5 Pro right now. Pretty bad website to be honest with you, but it works. It's super basic. You'd have to spend a lot more time just coding out, building out, but you can see here it does actually work. It has c created like a five-page website. You need to go back and forth with Gemini 2. 5 Pro. I would give that a five out of 10. All right, as a oneshot prompt. Let's run the same thing now using 03 directly. So, we're going to take the same prompt and then we're going to start a new task. We'll go inside the settings and switch over to 03. All right. Open AI 03. Make sure you don't select the mini model. You want to be using 03, which is the latest model here. And then we'll go inside here and we'll give exactly the same prompt to chat 03. See what we get back. And we're getting an error right there. That is not ideal, mate. What's going on here, sunshine? All right. Maybe we'll switch to Open AI directly. Just going to have Oh, look at that. Your organization must

be verified to use this model. All right. So, it's not working on Open Router. And it's not working. Let's see if we can. No, we can't. Open 3 is not working in Rico. That is so weird. Let me try a client instead. Not a great start there. Let's switch over to client. But again, okay, what's working? What's not here? So, Gemini 2. 5 Pro work perfectly inside Open Router. Let's test out and we'll just grab a new API key from Open Routter. If you ever need to know how to do that, just log into OpenAI, grab a key, and then you can create a key here. I'm just going to do that in the background in a sec. Run over to Visual Studio Code, plug the new API open router key here, hit done, and let's retry that. You kidding me? 03 is just not working on the API. That's crazy. All right, that settles it. So, five page websites, Gemini 2. 5 Pro. So, just to recap here, this is blowing my mind because I've never seen anything as objective as this. It's weird and feels biased, but it's not. So, test number one, the football simulator game 2. 5 Pro smashed it. Test number two, the car game Gemini 2. 5 Pro landing page Gemini Snow Day Gemini and five-page website open router wasn't working. 03 inside the playground wasn't working for me because I had to verify my business which means I can't actually start using the API. Massive problem. That is a full suite, my friends. house that Gemini has. Crazy stuff. I've never seen that before. I'm absolutely gobsmacked if that's an actual phrase. All right, I'm not going to do any more full website tests inside root code or anything like that because, you know, it doesn't work. Let's give it one more shot. I'm going to give this one more opportunity. 03. We're going to try and build a retro keyboard here. So, let's go back into 03. Feels pointless at this point, but I just want to be I want to give them a shot. cheeky chance. All right. So, we're going to say build a retro synth keyboard in HTML. We'll do the same inside Gemini and we'll see what we get back, peeps. Here's the code. Let's give it a cheeky whirl. I don't know why it didn't let me test that in canvas to be honest with you. But anyway, let's plug this in. So, this is the output from Chat GBT. It's not looking bad. It's not bad. Let's test this out. All right, it works perfectly. It work. It definitely makes a noise. Here is the single HTML file. Copy and paste it in the browser. Let me just delete those at the top. There it is looking pretty retro. We'll just delete the HTML code at the top. There we go. We got the retro simp. Not bad at all, my friends. Not bad at all. Let's check out Gemini now. I should have tested in canvas. Rookie error there. Let me build it again with the canvas enabled. It's got Yeah. So, let's see what we got here. So, we got the Retro Synth keyboard. It works perfectly. We can switch between waveforms and we can change the volume as well. So, that's working really nicely. Now, let's check out the Retro Synth. The main thing that I would say here is like the UI is pretty nice on Chat GT, but there's no volume. There's no switching between waveforms. The keyboard itself looks tiny on the preview, which is weird. Like all this black space. I'm going to have to say my friends retro synth keyboard. I'm going with Gemini 2. 5 Pro. All right. And the reason for that is it's got the waveform UI here. It's a much bigger keyboard preview. The you can switch on the volume. Again, 2. 5 Pro absolutely smashes it. So overall, you can see who the clear winner is. I'm going to stick with. But just over the last sort 30 days, Google was really stepped up. like Gemini 2. 5, sorry, Gemini Advanced Pro was never something on my radar up to about 30 days ago. So, you can see who wins. You can see who the winner is. It's unanimous. Absolutely unanimous. So, thanks so much for watching. If you want to get a community packed full of people focused on making money, saving time with AI, feel free to give that inside the AR profit boarding link in the comments description. Inside the classroom here, you're going to get all my best workflows that tried, tested, and proven, and it's way better. On top of that, also, you're going to get a crash course based on everything I've practically implemented into my business right here. So, you can steal all my same systems, my best SPs as well. There's tons of SPS in there. You can watch back previous Q& A call recordings, plus join the weekly Q& A calls like you can see on the weekly schedule here. And on top of that, you can post inside the community and add 670 members any questions that you have about AI that you're struggling with. On top of that, if you want to get a free one-to-one SEO strategy session that shows you how to get more leads, traffic, and sales, feel free to get that link in the comments description. And we'll show you how we take websites from zero to 145,000 business month and generate hundreds of thousands of dollars in sales and autopilot on this free SEO acceleration session. You're going to get a custom tailored game plan to generate more lead, sales, and profits based on what's working for us. And you'll also be able to ask any questions you have onetoone on the call. So feel free to get that link in the comments in description and

I will see you on the next one. Cheers. Cursor and Gemini 2. 5 Pro are absolutely insane. I actually just built this Tetris game in one single click. I'll show you exactly how to do it, but it works perfectly. So you can see here, for example, we've got all the pieces of Tetris puzzle. We can move them around. We can switch these back in over here. And this is just working beautifully. All right. It's incredibly easy to set this up using Gemini Pro. Now, I actually think from what I've seen on the tests and when I've been building with it, Gemini Pro seems to actually be better than Claude in many ways, Claude 3. 7 Sonic, but the difference is of course if you're coding directly with AI Studio and AI studio. google. com, then you can actually get free API. So you can code for free using this and then build whatever you want like games, tools, apps, websites. This is just one silly little example, but you can see how easy it is and the features functions perfectly. All right, so if you want to set something like this up, what you can do is you can download cursor and then inside cursor for example, literally all we said was like create a Tetris game right here. And then you can see it went off and coded it and created the logic. And then we've actually got the index html ready to go. And if you want to choose Gemini 2. 5 Pro experimental, you just click it down here. You can also switch between auto select. So cursor can switch between them. And you can also enable thinking, but for this project, I didn't even enable thinking. It was really simple and easy to do. And then it coded out the HTML over here. And if we open that up in Finder, so we'll open that up. And then if we grab a new Chrome, plug this in over here, you can see it's ready to go. If you actually want to host something like this, you can go over to Netifi. So we can log into Netifi over here, add a new site, import an existing project, grab the folder. All right, so you just grab that folder and then plug it into Netify Drop right here. Pretty simple stuff. Really easy to do. So, let's try and build some more stuff with this right now. Just going to open up new tab. So, a new chat over here. We've got Gemini 2. 5 Pro Experimental. And I'm going to say, okay, create a really sleek and interesting keyword research tool for SEO. Make the design the best it can be. All right. And then I'm going to say okay and promote these links with Alex for mozzy style sales copyrighting through app. And then we'll put the funnel links underneath like you can see and then all we need to do here is just hit send and that should start doing its magic. All right. Now you can use something like for example Visual Studio Code and then you can start from there using R code or client to build something similar but honestly it's going to run out of tries pretty quickly. That's the problem is like when you're building stuff out with the main API from Gemini it seems to hit the limit on requests really quickly. So for example, if we select Google Gemini Pro Experimental inside, for example, R code here, we'll test out the results side by side. So I'll take exactly the same prompt and we'll test the results. All right, so we're going to plug this into Visual Studio Code. We're using Google Gemini to build that out and we'll see what it comes back with in a second. All right, so you see here how there's an error. Super annoy. So we've hit the limits. it says resource exhausted and so that's the problem right now it seems like the API is just being rinsed and super hard to use whereas for example we've got the file ready to go so it's created the HTML etc to make this fully functional blah blah we'll accept all the files let's just see what that looks like so far so this on a new tab grab the index html and there we go keyword research tool ready to Let's press this now. Is it working? Yeah. Look at that. Perfect. All right. So, if you compare the difference here, like cursor is way better for coding with Gemini 2. 5 Pro. Obviously, this is mock data. So, you would want to insert like API integration. We can actually do that in a second. And then you can see on the page here, it's got links to our funnels. Right now, if you compare this versus what Root Code did. So, if we go back to Visual Studio Code, I am a big fan of R code actually. But you can see here how it's not going to go off and do stuff for us like we have to click run command. We have to mess around quite a bit. Whereas inside cursor, we're good to go on that, right? We just give it one prompt. It's like boom, here you go. And you've got that page ready to go. What we can try from here is setting up the API integration. I also like the fact that it's told us this is mock data, right? So the AI is not trying to

mislead us and tell us, oh yeah, this is ready to go, etc. It's just found mock data. So, what we're going to do now is we'll go back to cursor, open this chat up, and we're going to say from here, okay, create the back end API integration with chat GBT API, plus make the design more modern, interesting, and sleek. And now you can see it's beginning to code all of that out. So it seems a lot more autonomous whereas for example if we go back to root code here we have to keep clicking save and go ahead and all that sort of thing. The biggest difference though however of course is like cursor is like $20 a month for the pro plan like this but if you get better results you save time it's 100% worth it whereas for example root code it's free but it's going to take you more time and then you have to click save and run and everything else as you keep going through this. Also, I think this will run out of API requests, whereas inside cursor, it's going to start doing its magic pretty quickly. So, let's see what we got here. Next step is build back end, integrate the Open AI. Okay. And I'm going to say, okay, create this, but let the user insert their own API key into it. And so, like you can build your own sort of SAS tools, apps, or whatever you want using this process. Pretty simple and easy. So it's beginning to do its magic right here. You can see it's editing the code. Okay. So now it says now the user can enter their keyword and open API key. When they click find keywords now the browser will directly query the open AI API key and show the formatted text response from chat GPT in the results area. Remember the security implications of the front end only approach. If you use front end instead of creating like a whole backend server and obviously it's going to be a lot faster and easier to implement. So, what we're going to do is we're going to right click index. html. Bear in mind, like I'm not really a cursor expert, but it's pretty easy to set up, right? Like, it's quite intuitive. It's really easy and clean to set up. So, now we can insert the API key and we can plug in our keyword. So, I'm just going to grab from OpenAI an API key here. All right. So, if we go to platform. openai. com and I'll grab an API key. Just go to dashboard and then API keys over here. Hit create new secret key. Obviously, I'm going to delete this afterwards, but I'm just showing you as an example. And let's just check this actually works cuz sometimes you can build this stuff out with, for example, bolt, but bolt new and tools like that, like they often break when you're using the API key. So, let's see what we get here. We're going to click find keywords. Now, it's asking the AI. It looks like it's actually used the AI. So, it worked perfectly first time around, but you see how the UI is number one, not that nice. Number two, it should be generating the results really here. Like, we need to make it cleaner. So, I'm just going to take these actions and put it here and say make the UI cleaner, more modern, and better designed. Make it the best it can be for front end design, right? Then for the keyword results, make this feel like a beautiful sasttol. What else do we need to do here? So, we're going to go back here. Yeah, it's super bland right now the way it's designed. And also, there's like lots of space on the page. So much And it's way too bland. I'm also going to say also give me search volumes, keyword difficulty, etc. on keyword research the best you can possibly be. So the other cool thing here as well is like we can switch between API keys really quickly. So you know if we wanted to say use claw 3. 7 sonet max which is typically better for front end and usually more expensive but we're using this inside cursor so it should be a lot easier then we should be able to get better results. We can switch between them or we can just use think auto select and get it to do that. It's magic right there. But it's pretty good as well. So, let's see what we get back here. I also like the speed of this. Like, you can see how fast this is. Whereas, you know, quite often if you're using R code, recode has not even created the tool yet. We're still waiting for that to be done. This requires like a lot more button pressing and just going yes, approve, etc. Even when it's selected on auto approve. So, it's created this tool. So, this is the tool from Brew Code. If we hit run command here, let's see what we got. I actually prefer to be fair, the front end of Root Code for the tool is much nicer than the front end from Cursor, which is pretty interesting. But let's go back here. We'll type this in. See what we got. It's got a nice little

CTA there as well. And we didn't have to prompt it properly for that either. Let's go back to cursor. All right. So, it's promising me that this will be a lot nicer. Let's have a look. See what we got here. So, we open up the new version. Plug it in. Here we go. And we've got the keyword. Let's grab the API key. Hit find keywords. And it's generating over here. So, it does look better. It's not perfect, but this looks a lot nicer than it was previously. So, it says find winning keywords, AICO, find keywords, plugged in the API key. Works perfectly. It's also got some nice CTAs here. I just think we need to change the color scheme of the page, but it's looking decent. And then we've got the keywords and estimated volume and all that sort of thing. It also says found gold. Turn those keywords into cash. Learn the SEO Elite Circle blueprint at the bottom. So, we're beginning to get something that looks better, right? It's got a lot more functionality than it did previously. Like, this is the old version. It was pretty nasty, but this version is a lot nicer. So, let's go back now. And I'm going to say, okay, the design has to be more beautiful, sleek, interesting, and modern. Right now, the page background is just white, and it doesn't feel modern at all. I say, could you create something that's really sleek, like for example, the UI of Airbnb, something like that. Give you some examples. See if that improves it. We'll wait for that to generate, see what we get back. But you can see the evolution is pretty big between the first generation versus the generation that came after that. Looks a lot nicer with all these ratings and that sort of thing. Here we go. Open that up. I hope this looks a lot nicer. Let's see what we get here. It's just not that interesting, is it? So average. All right, here's what we're going to do. So I'm going to say use my brand colors. I say add like animations relevant SEO icons make the page fun interesting book 10. This is a 100 million SAS tool. I think part of the game here is like just figuring out what prompts are going to get the best out of this page. So, I'm going to run that again. Does seem insistent on just having a blank white screen, which doesn't make any sense at all. The images are looking better. I mean, the buttons here are looking better. Let's see what we get back. Now, in the meantime, I'm just going to open up a new window. And what you can also do if you go to your settings over here is you've got a bunch of new features, right? So, for example, if you go to MCP over here, you can actually add MCP servers inside Cursor as well. And then you have it looks like you have to create an MCP file for that to work. Also got a bunch of features here. Some new rules. So you can add like different rules inside here as well. And then you can select which keys you want to use. Now if you want to use your Google API key directly, you can just plug in here. Same for anthropic and open AI as well and Azure. So Azure will help you. I think you can for example like host that through the Azure API. And then there's also a bunch of new features here. I've not used this for a while, so that's why I'm saying, but you've got, for example, like the web search tool. You have all these sort of auto excepts or partial excepts. And yeah, it's definitely improved a lot since I last used cursor. So, let's go back here. We should have this ready to go. There we go. Now we're talking. Now we're talking, my friends. All right. So, let's try this again. Now, let me grab this API key. Plug it in here. We'll put, for example, SEO agency. Hit find keywords. Do these links work? Yeah, we're good to go on those. Beautiful. And then you can see on the results here, it's broken it down. So, we've got the keyword phase, SEO, agency, estimated volume, difficulty, CPC, andformational, right? So, whether it's like for example a commercial intent, navigational, informational, etc. And we're good to go on that. And then yeah, that is looking a lot nicer. I think that's actually probably ready to deploy. So what we're going to do from here is we shall let's go down here. So if we go into this review on Finder going to copy those files and then we'll create a new folder and we'll call this keyword tool. All right. And then I'll show you how to deploy this as well. So if we paste in those files over here. All right. And then we're going to go over to Netlify and I'll deploy this so that you can use it as well if you want to. All right. So

we'll import this. Try netify drop. We'll grab the folder that we just created. Just plug that in. Wait for that to upload. And then boom. Look at that. It's hosted on a subdomain. So we got that ready to go. It's hosted at netify. app. You can test out if you want to. You just copy that URL. I'll include it inside the AI profit boardroom as well along with an SP if you want to steal this from me. But essentially, that's how easy it is to create like a nice looking SAS tool that works with an API key using cursor and Gemini 2. 5 Pro, right? And also you can see how easy it was to build Tetris using this process. Like you can build apps, tools, games, whatever. And Gemini seems to be a lot better for building games out, right? Really simple and easy to do. So, thanks so much for watching. If you want to get the SP from today, feel free to get that inside the AI profit boardroom. This comes with all of my best workflows, tips, SPS, templates, agent templates, etc. And I'll include an SP inside the SAP's whole collection section over here for today's video on cursor if you want to build this stuff out. Also comes with a crash course, Q& A call recordings, weekly Q& A for me. So, you can jump on the weekly Q& A here and ask me any questions you have. And inside the community as well, you can post any questions you have. Then you've got 600 members dedicated to making more money and saving time with AI. So if you want to join the community, make sure you join actually now because we've only got until the end of March before the prices go up. So at the end of March, prices are going up. You want to lock in your legacy price right now before it goes and you can get access to that link inside the comments in description. Additionally, if you want to get a free SEO strategy session, feel free to get that. And on this free SEO strategy session, we'll show you how we take websites from zero to 145,000 vis and generate hundreds of thousands of dollars in sales on autopile. On this free link building acceleration session, you get free SEO domination plan. Discover the secret SEO link building. We answer any questions you have. You'll learn the best link building strategy for your website, plus how to 10x SEO traffic based on what's working for us and our happy clients like you can see right here. Feel free to get that link in the comments and the description. The new Google Gemini 2. 5 Pro update is absolutely insane. You can see it just got released a few hours ago. It's completely free to code with if you do what I'm about to show you. And I literally spent like 5 minutes testing out this new update. You can see I built AI powered keyword research tool, a CRM, and even a Roman marketplace simulator. You can see right here, which we can play kind of like a game, and it was super easy. I literally just gave it like two prompts, one shot, boom. I'll show you exactly how I built it. to access all this for free and test it head-to-head against Claude spoiler. One of them was not even close. And I'll show you exactly why that is in a minute. Plus, I'll walk you through setting it up with AI Studio, Visual Studio Code, and even show you the new Bolt DIY integration that works basically perfectly. And if you've been waiting for coding AI that actually delivers in one shot, one prompt, like you can see right here, then this is it. And let me show you how to start building with AI in just a few minutes. Let's go. Brand new update from Google Gemini 2. 5 Pro as you can see right here. I've been testing out today. Actually just created a full CRM like you can see and even a 3JS Roman marketplace simulation which is pretty crazy. We can zoom in and zoom out of that right there. And it looks absolutely banging to be honest with you. It looks really good. And so what we're going to be doing today is just testing out, seeing how it works, seeing what we can build with it, and seeing how powerful. already just from the look of feel from what I've built. I've literally just run these two tests in 5 minutes. It's absolutely rapid when it comes back to you. And if you're wondering, okay, what does this mean? What's all the new differences, etc. So, this is Google Gemini 2. 5 Pro Preview. And basically what it's focused on is even better coding performance. Now, I would already say that Gemini 2. 5 Pro so far has been goated in terms of performance, in terms of the outputs. It's getting better all the time and this is just another massive upgrade. So, if you want to get access to this for free, you can go over to aistic studio. google. com. Then you can select Gemini 2. 5 Pro preview in the top right here and you will see the May the 6 update right here. Okay, so this is available. It is free to use. I've not got my credit card details plugged in here. I think it does have some limits, but you can use it for free. And then if you have Gemini Advanced, you can also get access to this inside Gemini directly here and then preview stuff in the canvas. Tends to look better to be honest with you, but it looks really cool either way. You can see basically in terms of the web dev arena leaderboard, Gemini 2. 5 Pro now ranks number one. And I would honestly say they came out of nowhere this year. Like the tool was never that good. And then Gemini 2. 5 Pro came out and it was absolutely awesome. So you can see it's absolutely dominating the

web dev arena leaderboard from my own test. I would say it's definitely better than claw 3. 7 Sonic. It's also a lot faster and it's awesome for building stuff out. So you can also see for example here you can do some other cool stuff. So for example, you can actually video to code now. So it's got this video understanding and then you can use that to generate code as well. It's really easy to just code stuff out directly and we'll be testing this out in a second. All right. Now, if you want to see what other people are building with it, here's some examples from the actual Google team. So, they said, "Here's what our team is building in Canvas with Gemini 2. 5 Pro, the new edition today. " And you can see, for example, someone here created like a fun sound guessing game. We've got another one where someone built out a whole website for a onehop app for looking through books and stuff like that. Someone else built out this tool for mortgages and calculations and stuff. And here's another one. This was a garden bed planner, which is pretty crazy. It actually looks really cool. And additionally, we even have some games. So, for example, you see here, this is a game for learning like about chemistry, molecules, and stuff like that. And additionally, we have the McBTH line rehearsal app. So, you can build anything that you want. Any idea that you have, you can build. And it is absolutely amazing. For example, for building out this CRM dashboard, here's how I did it. This is how easy. It's absolutely silly. So, if you want to go and build stuff, you just go to new chat. Then you're going to select 2. 5 Pro. You don't need any coding experience now to build whatever you want. And you can see, for example, here what it said was build a CRM, right? Very simple prompt, not like a particularly difficult prompt at all. Then I said yes, because it was asking me some ideas and what we need to do, etc. And then from here, it was like, okay, great. Let's just go and build it out. And it literally built this out. I asked it to build it into HTML and canvas. And then finally from there it built something but it was not working when I was clicking through between these. So I actually said make sure that I can click through the tabs when I'm using it. When I click each section nothing happens. Please fix it. Boom. Shack lacker. That was fixed and corrected. Like you can see here's another example. So I said create a real world simulation of a marketplace in Roman times in canvas. And it created something that was very basic. If we actually open this up, you can see the previous generation. It's just like a flat sort of 2D simulator right here. Then I said, "Okay, make the people more realistic and make it 3DS so I can see the whole thing and move around. " And it was like, "Yeah, no problem, mate. " So it went off. It did that. And now you can see we can actually just use that right here. And I think the more iterations you do inside 2. 5 Pro, the more realistic you can make it look. But you can see how cool that is. For example, if we said maybe we want to for example add more detail, right? You could say add more detail, improve the market stalls, make it all feel more realistic, have more detail to the people, etc. Right? And then if you want to share this stuff, you can actually click on share over here. It will create a link for you. And this is just implemented on a subdomain inside gemini. google. com. So, if we click on confirm a few, we have the marketplace ready to preview right here. Really easy and simple to do. Super fast. Um, you know me, I've tested, for example, tools like Quen, Claude, and if it lets me down, I just show you on the live stream, right? But this one was actually super powerful. So, let's try this. I'm just going to say add more detail, improve market source, blah blah, and we'll see what we get back now and see if we can just take this to a whole new level. Now again if you want to use it inside AI studio you can do exactly the same thing. It doesn't seem to have like a canvas option. So you would have to for example just go inside the chat here use similar sort of prompts and then we can say okay building 3JS and we'll just put just HTML preview. Hit run and then we should be good to go on that. So it tends to take a little bit longer inside AR studio but you also get more control over it. So for example here you can check the token count. temperature. You can also switch between like code execution, function calling etc. You can even add Google search as well which is pretty cool. And additionally you've got like some advanced settings on the right hand side as well in terms of like output length and that sort of thing. Now if we go back we've got the Roman marketplace with a bit more detail, more buildings etc. It's added more detail to the marketplace stores. And if we look at the people, you can see they've just got a bit more detail to them. Now, if we keep going and we keep iterating on that, eventually we'll get something that looks really good. But as a first sort of what is this, a third draft, it's not bad at all. And it looks really cool. You just build out your own games. If you spent a whole day on this, you could

create something absolutely awesome to be fair. Someone asks, "What does temperature do? " So the temperature inside AR studio will just make it more creative, right? So it's a creativity allowed in the responses. If you increase the creativity, temperature, then it will be more creative in its outputs, but it's also more likely to break or to do something nonsensical as well. All right, so that's coding this out. Now we've got that good to go. Let's have a look and we'll just run that on the preview. So we're going to go to HTML code editor. Plug that in. We'll have a look at liveweave. com. tends to be better for previewing code, especially if it's something like this. And now we have the Roman market simulation like you can see. To be fair, I mean, I actually think the outputs were better inside AI Studio than they are inside Gemini Advanced cuz that's added more detail already. But yeah, you can see here the market stores, the people, etc. You could also animate that as well. So you could go back inside Gemini AR Studio, make the design more interesting and animated and lifelike, but I would say it's like the ultimate vibe coding tool right now because you can just keep going back and forth with it. Very quick to make changes. Now the other thing is here as well is if we pull this up and we compare it versus something like Claude. So let's do that. All right. And what we'll do is we'll just run exactly the same thing on both at the same time, right? To try and make it fair. So, if we go back to a new chapter here, we've got 2. 5 Pro experimental and then we've got Claude, which I would say is the second best option, right? So, you got two different options there. Again, this is for coding specifically like this is what Gemini is really good at. So, if we go into for example here and we'll test which one creates the best outputs. So, let's see what we got. Yeah, we can try with the simple 3D car simulator in 3GS. So, let's plug that into Gemini and we'll plug it into Claude. I'm going to give Claude the head start. So, we'll hit enter on Claude like and then Gemini and we'll see which one number one performs fastest and which one gives us the best outputs as well. I just need to select canvas as well inside Gemini advanced. Someone said compare it against Quen as well. Quen is just from what I've seen so far is so bad at Alberts. So the other thing here as well is like by default claw doesn't have thinking mode selected whereas Gemini 2. 5 Pro you can see it automatically goes straight into thinking mode before it starts coding anything out and you can see now it's coding out Claudius doing the same thing and we'll see which one comes up with the best outputs. So you can see here let's have a look. This is crazy. All right. So, they pretty much finished at the same time here. But what you can see is it's come back to us with this blue screen, my friends. Imagine if I was trying to do a tutorial on YouTube and I'm trying to show you some cool stuff that you can do with AI and then the tool that I pay £150 a month for doesn't give me anything. But the tool that I've got completely for free inside AI Studio works perfectly. That's a big difference here. And look at this. We're zooming around. We're having the time of our lives driving this 3JS car and it is working perfectly. Whereas, if you go over to Claude, nothing is happening, right? I'm pressing up and down. I'm pressing the arrows, etc. It says speed is 100 km an hour, but I can't see anything. And there's no preview here or anything, right? It's just it's absolutely that's what you get back from Claude. So you can see the difference here. Exactly the same prompt, but Gemini just gives us way better outputs, like 10 times better. And also, it's very easy to deploy this. We could add more detail to it as well. We can zoom around the map and we can build out as much detail as we want inside this project and then deploy it as well in a subdomain. but it's miles above the second best coding tool in the world. And I think until Claude releases like version 4, it's just not going to be able to compete. It's nowhere near the same level. Now, the other cool thing here as well is if we go over to open router, so I've already shown you how to code and vibe code with it inside, for example. By the way, this is the output. It looks better now. But yeah, so I've already shown you how to, for example, code inside Gemini Advanced. I've shown you how to get free access to Gemini 2. 5 Pro inside Google AI Studio. And if we take that iteration of what we created previously, let's see what we got back here. Yeah, look at that. It's just magical, isn't it? We got people floating around. We got people walking around the market store. We created that in one shot and it was

incredibly easy. Let's pull this up full screen for you. Like this is I don't know if anyone here used to play Mega Drive back in the day or whatever. This is probably better graphics than something like an old school Mega Drive. It's kind of like when the N64 came out. So that's what we're looking at here, peeps. He's super powerful, easy to create stuff, easy to build stuff. You don't even need to be a great prompter at this point, let alone need to code or understand what code is. It's incredibly easy. Now if we go over to open router we can also select models over here and we can start coding with the models for example if we go to this page I think we should have I want to make sure have we got open router on Google on have we got Google 2. 5 pro experimental on open router here we go all right by the way just make sure you filter to buy newest. I think Open Marita has changed that recently. So, if we click on here, we have the April model, but I can't see the latest model from Google. If we go on to the open router homepage here, yeah, it doesn't seem to be available on Open Router yet. But what you can do is if you go over to AI Studio and then inside here, you're going to get an API key, then create an API key. Once you've done that, then you can go over to something called Visual Studio Code. I have the full recipe inside the AI profit boardroom link in the comments description if you want like the instructions on how to set up. And then from here, if we go over to client, then select Google Gemini. Let's see if we've got this available to code with. Here we go. All right. Here's how to do it. Right. So if you go inside clin and then using 2. 5 pro, it's upgraded right here. So all you need to do is use 0325 in client and that automatically points to the newest version. So if we go to 325, let's do that now. Client inside the inst extensions 325 is the one that you want to select. Then you're going to plug in your API key that you got from AI Studio earlier. And then inside this section, just make sure you do have pro experimental setup. And then inside here, if we X off here, we'll go back to the AI profit boardroom. And we'll just grab a random prompt from here. So, let's see what we can do here. Grab that. Now, inside the custom instructions, I'm actually going to delete this stuff cuz I don't really use it that much. All right, let's hit save on that. And then we're going to plug this in. And I'm going to say generate a high converting landing page for my SEO agency. Just see what it pulls up. So this is using the latest version of Google Gemini 2. 5 Pro Experimental. And that was super easy to set up. And you can see here it actually gives you the API cost. So this is proof that you can code with it using the API for free, right? It's not costing you anything to use this before I start getting YouTube comments telling me it's not free. So it's created the HTML. Now it's creating the CSS. Let's see what else we can build with in the meantime. So, I'm going to say let's have a look what we got here. Really, you want to use it for coding. So, I'm just going to focus on coding related tools. Let's try a keyword research tool over here. Let's go inside Gemini. Now, plug that in. Always select canvas and then hit enter. What I'm going to do in the meantime is go back to AI Studio. We'll try and create something cool in here. So you can code like three different projects here with Gemini, with AI Studio, and then also with Visual Studio Code. And then whilst you're waiting for the code to generate, it's quite easy to flick between the tabs and just test different stuff out. See what else what other cool stuff you can build. Right. So I'm going to say, okay, 3JS SEO related game for my SEO agency focused on building back links. Make it fun and interesting like a 3D run a game for SEOs. Make the graphics awesome code in HTML. Then if you press command and enter that will just run the prompt so you can get the outputs. In the meantime, we've now got the tool back from the AI powered research tool. So it's created a keyword research tool that we can just plug in an API key and some ideas of like keywords we want to generate. Right? So for example, if we put dogs inside the niche and then inside the API key, I'm just going to grab that from AI Studio in the background. Basically like you can build your own custom tools. Now this is not guaranteed to work, right? Quite often when you build these types of tools where you plug in your own API key, it

can break, but let's see what we get back. So I'm going to copy and then plug this in. Hit start keyword research. That's finding the keywords for us here. And I'm going to say it's not using AI to generate this. I say don't use presets. Get the AI/ API to generate keyword ideas based on whatever niche I typed in. In the meantime, if go back to the chat now, that is already generated. It's ridiculously fast. It's so fast. This is silly. Now, let's go back to LiveWave and we'll test this out. We'll plug that bad boy in and we've created the SEO runner game. So, you see here, for example, you have to get tools, you have to avoid the spam, you increasing your DA, etc. And it works pretty nicely. All right, then we've got the penalty, and that messed us up. Now again, if you want to improve this, just go back to your code here and just say, "Okay, make this more colorful, add much more detail to the whole level. " SLRunner, make it feel more realistic. And I'm add a retro vibe to it as well. Make the background like an SEO traffic chart. And then we'll run that and we'll see what we get back in a sec. In the meantime, this has come back to us with the new code. And we can just keep flicking between everything. So we can v code like three things at the same time because we're waiting for all the different pieces to finish. Right? So if we hit start keyword research over here, boom, look at that. So it's actually worked now. Again, we just put one prompt there. Then we went back. Always just give it the feedback on exactly what you want or what the problem is. And then we've got the keywords ready to go right here. and it's generated that using AI, right? You could also say, okay, like tailor the outputs more to the topic, right? So, what is dogs? Obviously, not grammatically correct. And maybe some of these are like placeholder keywords as well. So, say okay, make sure you avoid placeholder keywords. Make sure the keywords are generated with AI. tailor the keyword ideas specifically to the topic I type in. Example, doc, you come up with keyword ideas like Google dog names using the AI and the API. Now, if we go back to Visual Studio Code, that is finished. Now we can hit run command and load this up. And we've got the oneshot page right here. There's parts of this that I just don't think that and I've always found this with Gemini particularly. I've never found it with claude so much, but when you use the API directly inside Visual Studio Code, quite often the outputs are not as nice as running it through the canvas inside Gemini. Right? So that's just one thing to bear in mind, but you can see how you can code stuff for free and it's pretty easy and simple to use right there. Now we go back to Gemini here. We've got the outputs again. This is the improved runner game. We'll go back to live weave over here and test it. Right. So you can see here now it's messed up the code a little bit and we can't see anything. We can't see any characters, etc. So if I was like going back, I would just ask it to improve that. And you can see that it's not perfect. Right? We've tested it to the limits, which is what I wanted to show you today. I don't want to show you like, oh, this is perfect if it's not. And you've got to see like the pros and the cons of this, but essentially, you can see how this works step by step. And then, additionally, if we go back to Gemini now, we've got the new version right here. So, let's plug in the API key again, see what we get back. And then we got some keyword ideas there. I still think that could be improved a lot. So, you can see the pros and the cons of this. like it can create the basics pretty quickly, but it does mess up a little bit. I don't know why it's simulating. It should be using the actual API call. So, I'm going to say make sure you do a live API call inside the tool, not a simulated one. The other thing you can do here is if you go to NA10. So if we log into NA10 here, right, you can actually build. I think I've actually cancelled my NA10 account, that's why. But you can actually build for free using NA10 as well, just using the API from Google as well. So it's quite easy to set up like agents. I've actually got some examples of how to do that inside the AI profit boardroom, right? So, if you want to learn how to use this stuff, you can actually just type in NA10 here, and then you'll see

all my best workflows, including like how to use stuff like Google Gemma and Google Gemini 2. 5 Pro for free inside NA10 to build whatever you want. Like you can see right here, we've got the full details like so. Today, we've covered all the ways that you can code with Gemini and what the tool does, how powerful it is, and what it can create, right? And also the limitations of this, where it gets stuck, and where it messes up. So, you know, it's not perfect. I've shown you what it can and can't do. And yeah, you get the point here. So, we've shown you Gemini. I've shown you AI Studio. We've looked at N8. We've also looked, for example, at Visual Studio Code. There's one more tool that I want to show you today, and that is Windsurf cuz I know you've probably heard the news. I think Chat Chubet is buying them for $3 billion, something crazy like that, right? So, if you want to use Windsurf, let me show you an example. Right. So, let's open up a new project here. I'll close that one. Then, we can just open up a new folder and we'll create a new one. We're going to open that up. I think they've just released actually. So, you can see it's downloading the update in the top right here. They've actually just released Windserve version 8. And then you've got these two options, right? So, you can use chat or you can use write. And then we're just going to restart this to update to the new version. By the way, these are all the new features inside Windsurf 8 as you can see. So, these are the new changes. I think this literally just came out. Yeah, May the 6th. So, it's a brand new update. And you can also see Gemini 2. 5 Pro is available down there. So, if we go to Windsurf now, click that. Let's load this bad boy up. Windsurf is totally failing on me. All right. Great. All right. At least you know then as well. Maybe you can't use like wind surf. I'm just being 100% transparent with you and you can see what works and what doesn't work today as well. So, thanks very much for watching. Some people are saying wind ser is more like wind crash. I would agree with you today. It definitely let me down part of the game. Usually, it's pretty good. Fair enough. They get purchased for 3 billion. You can't open it up. Doesn't make sense to me. So, thanks so much for watching. If you want to get access to the AI profit boardroom with all my best workflows, templates, SPS, etc., including how to code with all of the tools I've shown you, feel free to get that inside the AI profit boardroom inside the community as well. You can ask any questions you have. Also comes with weekly Q& A calls, and additionally all my best courses, agents, workflows, templates, and daily updated SPS. We update this bad boy every single day with new updates on all the latest stuff that's happening right now. Additionally, if you want to get a free one-to-one SEO strategy session that shows you how we take websites from zero to 145,000 visits month and generate hundreds of thousands of dollars on autopilot, feel free to get that link in the comments description and we'll show you exactly how to increase your SEO traffic, get more leads, traffic, and sales for free using SEO. One thing I want to try, I've totally forgotten about this, but I want to try it right now. So, let's give it a whirl, and that is bolt DIY. So, I'm going to shut down Windsurf. Shame on you, Windsurf. And then what we can do here is we'll grab the instructions for setting this up like you can see right here. And then we're just going to get this set up on terminal. All right. Once you got it set up, we can go to the local host. If you ever had any problems setting up with the SAP or whatever or GitHub, whatever, just plug it into Claude. It will save you a lot of time. It'll tell you exactly what the problem is. And then what you can do from here is we're going to select Google. And from here, we're going to select, let's see what we got here, 2. 5 Pro Experimental. Or you could use preview. And then if you ever want to code with Pro, you can just go inside here, select the model that you want. So this is a brand new model by the way. The it's created May the 7th, right? So it's pro preview that you need to select. Then from here we're just going to grab an API key from open router. So go to open router top right keys create key. Then we're going to plug in the API key here. And then from here let's test coding. So we'll just say build just for the sake of time build a simple SEO calculator. And there you go. All right. Super easy and simple to do. Paste in the article. paste in the keyword works perfectly first time round. You can use bolt DIY set it up locally code locally and to be honest actually works plus it's free whereas wind surf it let us down directly on the video today. That is basically how to use all of this stuff, how to code with anything. We've talked through Bolt, Gemini, AI Studio. We've also talked through Windsor, Visual Studio Code, and NA10 as well. If you want to get all my best recipes

templates, workflows, it's all inside the air profit boardroom. We update the SAPs daily. We're adding new updates all the time. Plus, the community is absolutely awesome. And if you join, you can actually become an affiliate if you want to make money that way as well. And it's super simple and easy to get set up. So, thanks very much for watching and I'll see you on the next one. Cheers.