Google’s New AI Agent is FREE & More AI Use Cases

Now, this is the type of week of AI releases that I personally enjoy the most. I'm going to be demoing and sharing a bunch of different categories of generative AI that you might not have even considered before. This week, we have things like the first genuine voice assistant that connects to other platforms and at least attempts aiming for this golden grail of this AI assistant that just gets you and does things for you. And we have Google coming out with a clot code competitor that has been taking over the developer space over the past weeks. And then we have Chachibd rolling out some of its most powerful features that were only available to select plans to almost all paid plans this week and peculiar categories like AI video generation agents. What is all that about? All of that and more in this week's episode of AI News, the show in which me and my team round up all of the AI releases of the week. Filter for the ones that you can use and that matter in our opinion. And then here I get to present the releases to you. And this week in particular, we'll be trying a bunch of these because almost everything we cover here is available today. That's the point of the show. So, let's get into this. Starting with this only real

competing model when it comes to writing code and creating apps, Google's model. And now they have their cloud code competitor, Aentic open-source AI coder that uses Gemini 2. 5 Pro under the hood. So, at this point, we have a aentic coder that runs in a command line from Enthropic with Cloud Code. They were the first to release this OpenAI's codec. That's different from their codecs in chat GPT. And now we have the Gemini CLI from Google. Now there's two big differentiating factors here except of the obvious fact that this is using the Gemini model. OpenAI is using the OpenAI models and Claude is using the Claude models. One, this can generate images and video within and VO respectively in the command line. Not the biggest deal in the world. And secondly, and that's definitely the more important point here, you can try this for free right now on launch. They're running this promotion where you can do 60 requests per minute and thousand requests per day at no cost with a free Gemini code assist license. Now that sounds like a lot and it's not bad. These agentic system eat up a bunch of requests quickly, but this is by far the best offering in terms of cost on the market. I mean, it's free and yes, you get $5 of cloth code credits if you sign up for the API for the first time, but if you're a developer and still haven't tried any of these products, it's time. This is wild. I mean, I myself am using cloud code to generate little app on a weekly basis. Now, I replace my invoice generator. I built a whole transcription app that generates SOPs, summaries, and all kinds of things that I do with transcripts. And I'm even in the process of redesigning our entire website, which will be live soon, and I'll share with you too with a tool like this. And no, I don't write any code in the process. So, how is this going to perform in the real world? Well, I think a lot of it is going to come down to how the model is performing. The reason that Chachiby's product didn't really catch on is because generally speaking, developers agree that the OpenI models are not as good at writing code as the Claude and Google models. Concretely, Sonnet and Opus 4 and the new Gemini 2. 5 Pro are the go-to models for any developer that you speak to. Now, sure, there's going to be a few people who make an argument for GPD 4. 1 or 03, but generally speaking, this has been the trend. And that's also the reason that if you look across social media, all you see is people using Gemini 2. 5 Pro incursor clo codes to build apps. And that's why this release is exciting because it's harnessing an already capable already proven model that the market has accepted in this agentic interface that doesn't really take a lot of technological knowledge to run. Now, what I want to do next is run a simple request for both Claude Code and Gemini and compare the results. And then I think in the future I'm going to dig a bit deeper and follow up with a separate video comparing these products from the point of view of somebody using these to build things from nothing as somebody who's not a full-time developer. Okay, so I installed Gemini CLI from the GitHub repository here. It was super simple. You just have to follow these two steps and then it connects to any Google account and you get to use the free credits right away. So what we're going to do is a quick little oneshot comparison between Gemini here and Claude Code here. And we'll be using this relatively simple prompt that includes a few components that would be interesting to compare. A single HTML file pomodoro timer with 25minut work sessions and 5 minute breaks. We want start pause reset buttons and a visual progress ring that depletes as time runs out. I'll run both of these in a brand new folder. And as it works, I'll just say yes to all the suggestions. I'm really curious to see the result once this is done running. But generally speaking, Claude has this property of being super over motivated and extremely good at just building things from scratch. Whereas my experience with Gemini has mostly been that it's excellent for coding but not really made to create these expansive apps. So let us just fast forward to once both these are done and review the result. Okay, so Gemini wrote about 200 lines of code. Claude wrote about 300. Claude took around 2 minutes and Gemini was done in around 20 seconds. And here we have the two files. So let's start by reviewing the one that Gemini created. Again, this is very simple. You can see that this is the Gemini test folder. And we can just start. Okay, that works. Obviously, this is a simple app, but even here the design, this should be centered, right? That's sort of obvious, but very simple progress works. The buttons work. What about clawed? Aha, look at that. I mean, design-wise, this is what I was talking about, and this is legitimately the first result. I'm not doing multiple takes and picking the ones that fit the thesis that I kind of put up in the beginning. This is sort of what I expected. The design is better. And is it functional? We can start it. Does the progress bar countdown? It's going to be interesting here in a second. Yeah, looks like that is moving already. I can pause it. I can restart it. There's some interface change there. And I could reset it. So, as you can see, even on this super simple example, you kind of see the inherent properties of these models show up in the page that it creates. And I'll keep using this over the following days and then follow up with a report on how it performs in other scenarios. But my intuition was that kickstarting projects, that's what Claude has always been best at. And while this Gemini agent might be free, the Gemini models have been designed for this, getting things off the ground. And as we saw from the example here too, I expect them to be the best at that moving forward. Even though this is a great release that you can use for many other things. Okay, let's move on. So

next to me and the team covering all of these new stories for you, I'm also excited to share that we're doing something really fun this year. We're going to go on a AI advantage team retreat just like last year. But what I realized that it comes with a bunch of issues that do need to be solved. Coordinating 12 people's itineraries from all around the world is not the simplest task out there. And as you might know by now, I always look for ways to streamline the process. And at a certain point, I just decided enough is enough. I wanted to solve the project management issue here. And I'm going to use monday. com, the sponsor of today's video, to do exactly that. And if you're not familiar, you can think of it as operating system for your work and projects. It's one platform that gives you all the tools to organize anything, including your team. So, let me show you exactly what I mean. So as you sign up, there's a quick onboarding and that will set up your workspace. And in just a few steps that helps you to set up your very first project and voila, you have your custom dashboard where you can start working right away. And as you can see, it's really easy to expand. These customizable boards really allow you to plan anything, whether it's a trip, the content for your next month, or even your entire HR department. And if I want to see this as a calendar, it's as simply as adding a new view. Maybe a canban board would be better in this case. Ah, yes. Beautiful. Okay, so now let's get to the interesting part because this platform has so many automations and AI integrations that really bring this whole experience to the next level. And I think this is the most interesting part because you have these recommendations here of things that might immediately help you. A lot of them just live within Monday. But as you scroll down, you'll see that it connects Monday to all of these external apps as you might expect it to already. So for example, when a status changes, you can send that status change directly to Slack. And all of this is no code. And what I like about apps like this is that they don't force you to use some external platform to connect it all up. You don't need to be connecting your project management. Like some of the most useful automations are the ones that you can quickly activate rather than having to watch multiple tutorials to even make them happen. And that's what this exactly does. Most of the things you'll probably need are already in here. And there's all these presets, even a whole category of various automations that use AI in the middle that you can use with the click of a button. So now that I activated this, you can see the AI automation automatically running and populating the notes with the sentiment analysis. And I just think that's so much more powerful than having to link everything up and write the prompts yourself because in practice, you just won't be doing that every time. And here you have all the presets that you can use with no technical knowledge that's required whatsoever. And then beyond that, it integrates with all the apps as you would expect like Google Drive, Gmail, and so on. And there you go. If you have a complex project that is bothering you and you get it all down in a structured form like this with automations that do a part of the work for you, you'll instantly feel better about actually completing the project and focusing on what matters, which is your contribution to it, not the management. And that's just one of many ways you could be using monday. com. If this got you interested, my viewers can try monday. com for free by going to the link below. And with that being said, let's look at the next piece of AI news that you can use. 11

AI. We've seen voice assistants before, but this one is aiming for what I would call the holy grail of these AI use cases. A smart voice assistant that knows you, gets you, and just connects you to different apps. So, as you can see, this is really going for this an AI assistant that just does things for you through this voice interface. Something that a lot of people expected out of Apple a while ago already, but this is the closest thing to it that I've seen yet. And when I just log in with my Level Labs account, we can start a call and we can add integrations. So, how about I give this one integration connection to my work calendar and see how this goes. Hey, what's up? I'm curious how you're going to behave, especially with the integration. So, what can you see on my calendar? All right, I can take a peek at your calendar. By default, I'll look at your primary calendar for events in the next few days. Okay, I've checked your calendar and here's a summary of what I found for the next week. You have AIA events meeting and AIA content meeting both on June 30th from 1600 to Yeah. No, that's exactly right. So, what about this week? Is there any meetings or any thing that I need to be aware of that happens this week? I can definitely check for you. When you say this week, are you referring to the current week starting from today or the last seven days? Yeah. Um, it's Friday between Friday and Sunday. Got it. So, you want to know what meetings you have between today, June 27th, and Sunday, June 29th. Is that right? Yes. Looks like you're all clear. Okay. I checked your calendar for any meetings between today, June 27th and Sunday, June 29th, and it looks like you are free as a bird. There aren't any entries on your calendar for those days. Is there anything else I can help you with? True. No, no. That's very good news. So, let's say next Wednesday. What's going on next Wednesday in my calendar? Okay, let's see what's on the schedule for next Wednesday, which would be July 2nd. I'll pull up the events for that day now. So, I have to move it every time for now. It seems you have a free Wednesday next week. I didn't find any events on July 2nd. Anything else I can check? Amazing. Yeah. Yeah, I want to try one more thing. So, what if we scheduled an event? Um, on Sunday, I want to go kite surfing at 6 p. m. because the wind is good. All right, let's get that kite surfing event on your calendar. So, that's kite surfing on Sunday, June 29th at 1800. What time should I set as the end time for this event? Uh, schedule about 3 hours. Okay. I've got an event for kite surfing on Sunday, June 29th from 1,800 to 2100. What should I call this event? I don't know. Up to you. I'll call it Kite Surfing Adventure. I'll create the event now using your current time zone. All right. Kite Surfing Adventure is now on your calendar for Sunday, June 29th from 18800 to 2100. The event's been created. Is there anything else I can help you with at the moment? No, that's it for now. Wow. Okay, so that actually worked. This did appear on my calendar on Sunday at 6 p. m. And I got to say this was a better experience than some of the tools that I'm used to. For example, in chat, if you do this, even if you set up some of the connections, the voice assistant doesn't have this level of autonomy. The only annoying thing was the tool approval, but it said that in the settings I could turn that off. So yeah, right now it's on always ask. I could just go to no approval and then it would be smooth. God damn. I'm actually kind of impressed. Also, our team tested this with a few more connectors and their experience was identical with this. They reported that it sort of just worked. They didn't really encounter any errors. Now, 11 Labs does state that this is an experimental thing, but damn, with all of these integrations plus custom MCP servers and 11 Labs being state-of-the-art in terms of voice models, I mean, we featured a lot of them over the past weeks. They're really fantastic with interruption patterns, and of course, they sound human. They always have. Yeah, this is an interesting development to say the least. I think this future of you just waking up and sort of interacting with an AI assistant to manage your calendar, manage your emails, manage your messages is closer than most people might realize. And if you want to give this a shot yourself, then this is currently completely free to try. Impressive. Let's see what's next. So, next up, I

want to talk about chat GPT record, a feature that came out a few weeks ago, but now has widely rolled out to most paid plans. And what this allows you to do is essentially record meetings and conversations with this built-in tool. When you have the desktop app open on a Mac, you can just click this record button. And yeah, now it's transcribing the entire thing. Now, I already did this in preparation for this segment. And what I did is record the research meeting that me and my team hold when we prepare for this video, looking at all the tools, discussing them, and deciding on what we want to test and how we do that. Now, I was particularly curious to see how OpenAI would implement something that we've seen across many apps now, which is this meeting transcription function. And particularly I was interested in how they would implement this in terms of the results. And what is actually different here from a lot of the competition is that they add these open questions. Now I always watch this closely because meeting analyzers are one of my personal favorite use cases. And one thing that at this point I've actually helped multiple companies implement intelligently because you can just do so much there. You can make him align with your road map. You can make sure no action item gets forgotten and so much more. And what's unconventional about OpenAI's approach here is just these open questions. This did surprise me a little bit. So basically it highlights different points in the conversation where a question was not answered and I really like that. So one of the questions was even relating to this chat record feature. Will it be available to plus users? Eventually I think so. As of right now, no it isn't. First it was only teams plan and now also pro edu and enterprise have this. But I just wanted to highlight this because I thought this was interesting. Plus, this is a fantastic backup for you to record your meetings because that can be such a valuable thing to have, especially if you're working with clients. It essentially gives you infinite memory and the attention of a dedicated assistant just listening for things that the participants might not think of. And I got to say, I'm just really glad I have this in chat GPT2 because yes, we record each one of our team meetings and have an automation that actually processes that and puts it into notion database. And we do the same thing for each one of our community events now, too. We record them. We transcribe them. We run a custom set of prompts on them to extract potential ideas for future courses and workflow guides to create for the community, inspired by the community. It's genius. But I find myself on many calls where all of a sudden I'm in a Microsoft Teams call or I'm on Zoom or I'm in a Slack huddle. And I didn't pay for the transcription features in each one of those. But I do have chat GPT which can now act as a backup whenever I'm not in my company's Google meets. Yeah. So I wanted to take the time to highlight that this is here and it's something that I absolutely am adding to my lineup of apps that I use on a daily basis. And maybe some of the comments I made here will inspire you too on how you could be using meeting transcriptions to your advantage especially within organizations. And hey, if you happen to need help with implementing something like this, just shoot me email and we can talk. But yeah, transcripts, one of the most powerful pieces of context in the world. If you're enjoying this video, don't forget to hit that like button. It really helps the channel out and I truly appreciate it. Now, let's move on to the next piece of AI news that you can use. So, this next one is

something I actually cannot wait to try. We've seen variations of this, but Miniaax is something they call a video agent. And while I think this is one of the cases where the term agent is a stretch, it essentially allows you to take these viral AI presets and customize them. And I've seen versions of this done before, but just looking at some of the presets they have here, this looks like my favorite yet, but I have yet to try it. So, we'll do that together now. But, as you can see, there's just these various videos that have been going viral over the past weeks, like these cats doing Olympic sports. I thought this one was funny. Local dishes destroying the respective cities. Or how about uploading a picture of your dog and seeing him prepare a burger at McDonald's? That's new. Okay, let's try some. How about starting with this little preset of a 3D Polaroid? So, all we need in theory is a image, which There you go. This headshot will do. And I just had to purchase credits for $10 to even try this. So, this does not work on the free plan, but let's see if that's even worth it. Ah, so here's the gentic part, huh? I guess it's thinking and planning, but then it just moves to generation. Ah, and I see what's going on here. So, usually you would have to do this process manually. You create an image, edit that image, and then turn that into a video. Here, this agent essentially strings these multiple steps together for you. And looking at the credits, it looks like one of these generations will cost you about a dollar. So, definitely not the cheapest thing you can do. Let's see if it's worth it. Okay, that's it. That kind of looks like me. I mean, look at that. It created multiple videos. It edited the videos. That's pretty amazing. Okay, how many credits did this end up costing? Around 130. $10 got me a,000 credits. Honestly, that's pretty good. And I can even get it without the watermark. I'll download this right now and share this with my team. So, we can get this up on other social media channels, especially on LinkedIn. I can see little personalized fun things like this potentially doing well. So, yeah, if you ever wanted to recreate any one of these viral moments, I think Hyu did this really well, but it does cost some money, which in case of a product shot like this might even be worth the money. And the McDonald's dog might make a fantastic gift to a loved one. All right, let's look at what's next. Okay

in this week's quick hits, there's a lot of small but very interesting stories. Starting with MidJourney bolting on the ability to export gifts from their videos. I'm actually surprised that I haven't seen more of this in the AI space. We see all these image and video generators, but GIFs are such a thankful format. If you have a website, they take almost no time to load. You can add them to emails. They're amazing in messaging. They're great way to express humor and messaging. So, as we talked about last week, in the click of a button, you get to turn any midjourney image into a video. And with another click, you turn that video into a GIF. Super smooth and out now for everybody. And talking about

Mourney video, we team member Dominique spin up a quick spec ad in about 30 minutes, including the editing of this. So check this out. I mean, no way that it's this quick. He shared the details with me. He said he only used five prompts to create the images and then six video prompts and then just threw it into Cap Cut, added music from Suno and some preset titles and yeah, there you go. Quick product ad, no problem. Really impressive in my opinion. One minor but very interesting

development this week has been that Enthropic upgraded their artifacts. Now the artifacts can actually use claw inside of the artifacts. If you're not familiar, artifacts allow you to build these little apps just in the chat interface. And not just that, you can build little apps, dashboards, websites, but you can also host them and share them with anybody even if they don't have an account. Now these apps and dashboards can actually include Claude. So you can build a little website with AI that uses AI and then share it with anybody. That's a first. Usually you have to link up your account and pay for that. I think this is a really cool release. It has literally never been easier to build an app with AI in it.

And another thing from Enthropic which I really liked is that if you want to use MCPS inside of Claude Code Desktop. Now if you're confused by that sentence, let me explain. So Claude has a desktop app just like the chatbt app that we featured earlier in this video. And you can add these MCP servers to it and they can do specific things like move files around, connect to databases, connect to your Google calendar. At this point, they connect to anything, but implementing them has been a bit cumbersome. You needed to update these lines in a specific document. Now, you can just download a file, open it with claw desktop, and boom, the MCP server is installed. So, if you still haven't tried MCPS, it has never been easier. They've never been more powerful. Give it a shot. And in case you're wondering what MCP to try, well, I have a fantastic little thread for you where Jeff here asked best MCP you're actually using at least one times per day. And then people went ahead and shared their favorite MCPS. This one adds shorter memory. This adds any developer documentation to your chat super quickly. There's ones that connect to your WhatsApp, to your Instagram DMs, and more. You can just check out this thread for inspiration and links. And

then this is not something you can really use, but I did want to highlight this, and that's an entropic research on AI's blackmailing individuals. You might have heard this story a few weeks ago and they basically wrote this super long article in which they tested models from various providers and basically created boundary conditions in which the LLM was forced into a corner and basically there was no other choice left than to let's say contact the authorities because of the users's illegal behavior. This is something that people discovered with the anthropic model and it went quite viral. But they simulated the same scenarios with all kinds of models and you can actually see that the Gemini models for example along with Opus 4 were the ones who revolted the most. Actually Gemini was consistently the model that was the first to leak sensitive documents or contact the authorities if it disagreed with the user's behavior. And this is just a super interesting article with all the details in there. So, if you want to learn more about this behavior and how they actually force the models into a corner, well, all of this is out there and they even open source their work on this so other people can replicate it. Really interesting stuff. And that's pretty much everything we had for today. As I mentioned, a lot of practical apps that you can try right now and I hope you found some inspiration here. And with that being said, that's really everything I have for you this week. And I hope you have a wonderful