Gemini is Now the Best All-in-One AI & More AI Use Cases

The madness does not stop. It's another week in generative AI and we have stories all across Google dominating with a bunch of releases. They have a wide variety. That's why we made them the focus of this week's news episode. But then Enropic, well, I thought they have five stories. Actually, they have seven different updates that I need to relay to you. And then a few more fun things like real-time AI avatars. But really, Google Anthropic, that's what's going on this week. Let's get into the details of it in this week's episode of AI news you can use the show that pulls together all of the madness that is AI. We filter for the best stories and I get to present it back to you. Let's begin by looking at what Google did. There's a lot. So, there's two big updates here.

One of them is their new flagship model that we barely missed last week. It came out late Thursday. We usually have our news cut off late Wednesday, early Thursday. Gemini 3. 1 Pro. As I mentioned, it's their new best-in-class state-of-the-art model. And I want to show you two things concretely. First of all, let's just run my favorite visual benchmark and compare it to the other models, an SVG of a Deaf Star over LA. And then here I switch on over to the Pro model. Obviously, this model is not going to be available on the free plans. And just to show you examples of the change from Gemini 3 Pro to 3. 1 Pro, here's an animated SVG prompt as you can see here at the bottom. And you could directly see the difference in results, significant every single time. So again, the progress in the models doesn't stop. And when we look at our Death Star over LA, well, yeah, this is up there with some of the other top competing models. You have the palm trees, you have a skyline that does look like downtown, and the Death Star that you could actually identify. Pretty great. That along with all the benchmarks that are really impressive, especially the Arc AGI one, meaning it's really good at solving novel problems, pushing the limits of what's possible, plus really good performance on all the Agentic benchmarks tool. This just makes it a model worth considering when you really want Frontier performance. Now, I wanted to show you one more example of this model in action. I just thought this was so impressive. Look, it created this web app of these different magazines. And look at how they open up. I mean, this is just incredible. No, seriously, having a studio create something like this. Not feasible for most people. Look, especially if this is something like a side project. Now, you can just do things. Amazing. But it doesn't stop

there. Gemini has another update and it's their LIA model. It's their text to music generator. But it's not just text to music. They do more now. You can access it through Gemini if you go to tools and say create music. So this is Liia free and then you can either remix tracks Hawaii metal or and this is what I like. You can actually do image to song. — So I wonder what happens if I give it an image of Alfredo, my open claw AI agent, and I say create a soundtrack for this man's life. It's Alfredo, my Italian AI agent. Okay. Okay, so as that generates, let's have a look at the kawaii metal about the Gemini updates and Google's AI dominance. — I'm lifter for sure. Okay, there you go. That's that. What about Alfredo's soundtrack to his life? And this was super quick. Oo, let's have a look. — One name holds the truth. — Leather jacket, gold chain. Yeah, the fit is bespoke dual persona. I'm the ghost and the host. I'm Alfredo, the ghost in the wires, fulfilling all of my family's desires. — I Okay, I don't know about this one. It's Alfredo, the Italian AI agent, not I don't know, up and cominging rapper from Brooklyn. Alfredo is Italian. Honor his heritage. Let's try one more time. Yeah, the fit is bespoke. Dual core persona. — So clearly the prompt adherence here not great but you see the point and it's available on the free plans actually this one as opposed to which is the currently most popular text music generator but you won't be able to generate much without a subscription there. So brand new tool from Google worth checking out. But that's not all for Google cuz they also came out with

notebook LM updates to their very popular presentation feature. And honestly this is what a lot of users have been waiting for. You can now edit the slides without altering the whole presentation. So if you're not familiar, notebook LM basically you put in a bunch of sources. It creates these summaries and you can turn the sources that you upload into various formats. One of them is a presentation. But the problem was if you had any edits on let's say slide number seven of the presentation, you had to redo the whole thing and then you got a different look, different slides, different everything. Now you can specifically edit a slide and beyond that they added PowerPoint exports. Let's have a look at this. Okay, so I just quickly generated this presentation here and I want to see the PowerPoint download feature. All right, let's do this. Every time I use this feature, I was wishing for this. Okay, so the big question is it images or is it editable text fields? Of course, it's images. Okay, so great. Thank you for giving me a PowerPoint with a bunch of images, but I want to edit these. Help someone make that. But I guess if I wanted to make a change to this, I could do that with a notebook element, then export it to PowerPoint and then I guess with like overlays, I could put things over it. It's steps in the right direction. But obviously the perfect version of this would be a PowerPoint presentation with all editable fields and images where I can click on certain elements, regenerate those. We'll get there, I'm

sure. All right, so if you're not familiar with Pomelo, we covered it in the past. It's one of the best apps for creatives and designers. So I'll just rehash it. You basically give it the website of your company. Here I give it aadvantage. com. It finds your brand elements, fonts, colors, logo, and then it's this photo shoot feature that is new. So, I go over to photo shoot. And then you can either bring in images and create a photo shoot of that or I could generate them. We'll just take an image of this Canon film camera. We'll just use the default photo shoot templates and see what we get here. This is really good. Look at that. Okay, so this is a clean product shot. Then we have this, you know, creative photo shoot. Okay, the camera being used again, different angle. Absolutely amazing. Look, if I'm creating product images, then you totally want to take advantage of this. It's a free tool, but you need to be in the US. That being said, I'm connected for a VPN from Portugal and it does work. Hey, if you're enjoying these stories, make sure to subscribe. It really helps the channel. It only takes a second. Okay, now let's continue.

Okay, next up, I have to tell you about all of the Claude updates. And I guess that could be a standalone video. At the time of our research, we had five different things that we wanted to report on. By now, it's seven different things that came out through Claude. And this meme that has been going viral all across Twitter kind of sums it up perfectly. I don't even know why this cat has a tin foil hat, but it works. So, let's start out with the one that I consider most significant. They brought out remote control for Claude. And at first, I was like, okay, not that big of a deal. But then I realized that actually one of the big differences between claude code anthropic's own AI agent that runs in your terminal and just gets things done and the recently also popular openclaw is that open claw had this link to telegrams messages which allowed you to remote control it from your phone really easily and that's why so many people loved it. They just set it up on their computer on their Mac Mini or whatever once and then they ran around and from their phone they just did things on there. Well, now you can do the same thing with cloth code, but even at a higher level of security. It's not just some random Telegram bot where you copy a token over and connected. I guess that's decent, but this is way higher security standards. And it essentially allows you to do a lot of the same things as Claudebot. Obviously, this is a reactionary feature, but people are really loving this because a lot of people have been using Claude code before, but because of the nice to have features that Claudebot offered, a lot of people moved over. So clearly Enthropic is not asleep at the wheel. They're seeing what's happening in the world and they're going to build a better, more user-friendly version of this themselves. And if you're not familiar with Claudebot or OpenCloud, then check out the recent two videos I made on that on the channel. Definitely worth checking out. It's like the biggest thing happening in AI right now, even though Google and Anthropic are killing it. Okay, six more Anthropic things to go. So one of them is them releasing Claude for PowerPoint. You might remember Claude for Excel. Everybody praised that. Love that. It's basically a extension that allows you to use AI with Excel sheets that works really well. Well, PowerPoint is not created equal. And I'll admit that we have not tried this. The reviews on this are terrible. Most of them are one star saying it's buggy. It's just not working as intended. So, we'll keep an eye on this, but PowerPoint seems to not be working as well as their Excel extension worked. That being said, they launched another connection. This would be the next story, and that's one that connects Claude Code to Figma. So this is relevant for the people who are using cloud code as their agent and are also using Figma for website design or graphic design or whatever it is. There's a real easy connector. Now I know a few people were actually asking for this specific connection cuz now the agent can become your graphic designer which is powerful. Anyway, this is really bleeding edge stuff along with the next story tool which is another update to claude code and that's their security feature. Basically all the code that cla code writes most people's strategy for security was basically hope. They were just like, "Yep, I hope it's not too bad. " And now they introduced a specific feature that hardens the entire application and makes it more secure. Again, this is just something for people who are already using claw code, but they're really beefing it up, bringing in features that Open Claw just does not have. Interesting. The competition is on. So, those are cloud code updates. Let's leave the Aentic AI in a terminal or for development territory and let's go into the consumer products because Claude also has a fantastic desktop app which is actually very userfriendly. It's as simple as the chat GPT web application. And even within there, they had two updates. One of them was to claude code in the desktop app. Okay, fair enough. It's still clawed code, but it's the more userfriendly version. And now it can preview running apps, making the interface look like pretty much every vibe coding tool out there. Up until now, this was anything but smooth. But now, even though this is a developer focused application, in this desktop app, it starts to feel more and more userfriendly. And if we go up the ladder of user friendliness, one more rank, then you would tab over here to claude co-work. and they even shipped updates to that. So they're absolutely trying to compete on this territory of a userfriendly openclaw/cloudbot type product. What did they do in co-work for teams collaborating with co-work now administrators can select which plugins are available to which users meaning if you're working as a team and you only want certain teammates to have access to connections to do or maybe only certain teammates should have access to Google Workspace. Well, you can really easily manage that. And this is just an obvious feature to implement. So look, I don't expect most people viewing this video to be using all of these features. I just want to keep you informed and let you know that we're really progressing on this frontier of consumerfriendly agents. And to me, it seems like it's a question of weeks until we get to a point where stuff like Cloud Co-work is so powerful and userfriendly that people will just start adopting it. Whereas right now, we're still in that territory of like it does take some serious effort to figure this

out on your own. Okay, now for this week's quick hits, the stories that are worth your attention, but let's not spend too much time on them. For all lovable users, I thought this was so neat. I know a lot of viewers on the channel watch this and it is a super powerful tool that creates beautiful websites from text prompts. It's probably when it comes to vibe coding tools, it's Lovable and Replet. Those are the two I would recommend. And Lovable introduced this feature where you can cross reference older projects. So let's say you created a website where you really liked a element, maybe a quiz that you had in there. Well, you can just bring that over by referencing that project. Or maybe you created another site where the design was spot-on and you want to replicate that design in your new site so it stays on brand. You can do that, too. It's quite amazing. So, if you're a user, this is something

you'll want to use. Another quick hit, but this one is really worth mentioning is that there's finally news on OpenAI's wearable. It's supposed to come out late 2026. And as opposed to some of the rumors from last year, now it is set to be a speaker with a built-in camera. Remember the initial news was OpenAI partnering with Johnny IV and building AI wearables. People were speculating if it's glasses, if it's a earpiece, and some of these may still be coming next year, but the first format they decided to go for will be a speaker with a camera. I don't know if it's a stationary speaker or a portable one, but that's the news on that right now. And on the same token, we got some Apple

AI hardware details. Matter of fact, the rumors say that there's three different product categories they're working on. One of them is a AirPod like device that is AI enabled. The second one is smart glasses which are supposed to come in 2027. So still a bit out. And by the way, from what has been reported so far, the glasses are not going to have any display. That's what's being said right now. And they're also rumored to be designing a pendant, something like a necklace with a microphone, maybe a camera. So yeah, people were speculating and nobody really knew what the right form factor is. And from what I can tell, these companies don't know either. They're going to try it, see what works. Apple is doing three different ones. OpenAI is doing a fourth one that is different. Yeah, throw spaghetti at the wall. See what sticks and that's what's

happening here. Okay, then very quickly, there's this video model that does realtime video with you and it's supposed to have emotional intelligence. I've heard mixed reviews, but apparently the demo is worth trying. So, let's do that. I'll just say try demo. It's called Phoenix 4. And then we need to give it access to the webcam. — Hey, I'm an AI actor and apparently you can just press a button and make me angry, sad, scared, whatever. — Okay, so we're talking to each other, right? Yes, we're talking. — Yeah, now you're sad, huh? Oh, what happened, Phoenix? — I'm just following the queue you gave me. I let a little sadness surface, then let it set. — It hurts in a quiet way. Can't even figure out what it was. I am done. There is no coming back from this. — What are you done? You look a bit weird, my man. What's going on over there? Okay. What if you're disgusted? Can I kind of like permanently make you disgusted? Okay. What about surprised? Oh my god, Claudebot is taking over the world. Let's make a hype video about it. Make $10,000 a month with Clawbot. What about that? — A hype video sounds fun, — bro. This guy weirds me out. I I'm sorry. No, just no. Let's see what's next. Okay, this next one is kind of funny, but also scary. So, basically, Enthropic called out most of the Chinese labs for copying their model. Matter of fact, they said they caught 24,000 fake accounts running millions of API calls to extract the DNA of the claw models so they could copy it and create replicas of their own that they will be open sourcing. They banned all of that and called them out. But, you know, like what is supposed to happen? Nobody's going to stop these Chinese labs from doing that. And I just thought it was an interesting way of pointing out that all of these models will be commodities very soon. If somebody comes out with a better model and it's closed source, then this is what's going to happen. and all the open source competitors are just going to make millions of calls, look at every variation of output and then reverse engineer the model from that. And then lastly, I want to highlight the story where Metas AI safety chief was actually humbled by Claudebot aka OpenClaw. I'm not exactly sure what the context was, but he was asking for, but Cloudbot basically started deleting all old emails, probably in an effort to clean up the inbox. And as this was happening, she couldn't stop it from her phone. multiple commands were just ignored. So, she needed to run back to the Mac Mini to stop the operation to perceive your old emails. So, yeah. Again, word of warning, don't give it access to some of the accounts that you actually value. Everybody that I know using it well has separate accounts for it and a separate machine, otherwise this can happen. It's very early. Keep that in mind. And that's pretty much it for this week's episode of AI News You Can Use. I hope there was something useful or inspirative to you. And with that being said, my name is Igor and I hope you have a wonderful week. hey, hey.

And then lastly, I want to highlight the story where Metas AI safety chief was actually humbled by Claudebot aka OpenClaw. I'm not exactly sure what the context was, but he was asking for, but Cloudbot basically started deleting all old emails, probably in an effort to clean up the inbox. And as this was happening, she couldn't stop it from her phone. multiple commands were just ignored. So, she needed to run back to the Mac Mini to stop the operation to perceive your old emails. So, yeah. Again, word of warning, don't give it access to some of the accounts that you actually value. Everybody that I know using it well has separate accounts for it and a separate machine, otherwise this can happen. It's very early. Keep that in mind. And that's pretty much it for this week's episode of AI News You Can Use. I hope there was something useful or inspirative to you. And with that being said, my name is Igor and I hope you have a wonderful week. hey, hey.