Google's Push for AI Dominance & More AI News You Can Use

Welcome to another week in AI. Between Google releasing a firehose of new releases, Anthropic lifting the ceiling on what's possible, and best-in-class open-source models from China, we have a lot of ground to cover in this week's episode of AI News You Can Use, the show that pulls together all AI releases, we filter for the ones that matter, and then I personally get to present it back to you. So let's start with the story that made a lot of headlines that I created a dedicated video on because I wanted to talk about it. Claude Mythos. Their flagship model that is too dangerous to be released. Obviously a PR move, but also obviously super impressive benchmarks. In short, my take is I think this is going to be as good as we expect because Claude has a track record of delivering the best models. But what warranted that discussion in the separate video was the second and third order consequences of us getting an agentic model that is just head and shoulders above everything we have now. So that's interesting. And we'll talk about that more once it comes out. But there's a few more Anthropic-related things that we need to talk about here. First of all, last Sunday, they cut off the ability to use your Claude subscription with OpenClaw or other agents. In other words, people who were previously paying $200 a month to play around there now will need to pay $1,000, maybe even up to $3,000 a month depending on the usage. For most people and what they're doing with those systems, that's not worth it. For me as a content creator who's been tinkering and developing more and more things that work there, there's actually a bunch of great use cases for content creation concretely that, yeah, even the increased price tag is well worth it. Also, I want to course correct on one thing I said last week. It was a story that I caught very last second and I misreported it to you. So my bad for that. It's about the source code leak of Claude Code. I said that it was an April Fool's joke. It was also a Claude Code leak that was an April Fool's joke. So actually it wasn't an April Fool's joke. It was real and there were some interesting insights from it. Nothing groundbreaking. I want to highlight one interesting thing that happened though. So if you're not familiar, Claude Code, which is Anthropic's agentic product, basically leaked. And users took the code and looked into it. What were they doing? How do they handle memories and more? And one thing they found is that Claude Code had a dreaming function where basically every day it stored its experiences in certain files that it then later would recall, much like a human does when we sleep. Great idea, right? Cross-session memory. Fantastic. And keep working with it and it remembers what you were working on yesterday. And within a few days, OpenClaw, which obviously is the open source competitor to Claude Code, added a new feature. Dreaming. So now if you use OpenGLO, it can handle memory in a similar way. And I think that's just so funny because this is going to keep happening. You know, you're going to see it in today's episode. We get new apps from Google, we get new completely open models from China that are almost as good as some of the best models available today. And I just thought it was so funny that one of these big tech companies is developing a proprietary system that is closed. And then a leak happens. And within a few days, the open source alternative implements one of its best features. I think we can expect that moving forward. It's gonna be an interesting time. Google, oh boy, did they feel like releasing a bunch of stuff to become a part of this week's AI hype. Let's have a look at that. So first of all, another interesting thing that I couldn't get access to yet from Google, but if you're on the Ultra plan, the $250 plan, and are in the US, select users are getting this already, and it's called AI Inbox by Gmail. It's Google's attempt to build an email assistant that gets you to inbox zero automatically. Look, they shared this screenshot, and I think this is one of the things that 9 out of 10 people would love to have. An AI agent that helps them manage their inbox in a contextual way where it knows them, knows the previous interactions. So from what I can see here, this doesn't focus on actually replying to emails. It's more like prioritizing the emails that you get and creating daily personalized briefings. Mind you, this will be just step one. I mean, they're slowly rolling it out just to the highest tier customers. Within 6 months, everybody's gonna have this on an affordable plan. And there's gonna be more proactive versions of this too, of course. And my Gmail agent is gonna talk to your Gmail agent to figure out and our Google Calendars have time slots so that we can meet in person and talk about AI. Utopia. If you're an Ultra account in the US, check your account. You might have this already. Okay, the next we have an application that went pretty unnoticed across social media, but I thought it was really worth mentioning. I don't know about you, but I wasn't familiar with Google Vids. It's a very simple video editing software, but even if you're not into that, they're using that as a vehicle for all these different applications that we've been watching over the past few months. Video generation model, music generation model, they're all putting it into here. For example, if I want to start out by creating an AI avatar for my video, I can do that. Now, I bring that up not because AI video is like the biggest and hottest topic right now, but because I love to keep an eye on how these apps are evolving every week, as you might know. And as I pointed out before, we're really seeing these bundling, and this is a great example of that. It's the same experience as before, but now I don't have to bring it over into video editing software. All right, guys, welcome back to the channel. I'm here in Soho, and we need to talk about the latest updates in generative AI. Damn, this guy's background looks better than mine. And if you're in a pro account, you can even create songs right in here. This is so easy to use and part of the Google Suite that if you ever needed to create videos with AI avatars, this does feel like the most frictionless way to do that today. And if you wanted, pro tip, you could start with a slideshow that you have, just add a voiceover and a custom song right in here. You don't even have to touch the AI avatar stuff. Additionally, it also has avatars feature where you can pick one of these avatars here or you can create a custom one. For now, it does not have the ability to upload your own picture. Just a question of time though. The voice you select makes a difference in how viewers respond to your video. Hey, what's up YouTube? Hey, look at that. That's pretty good. You can now easily add other videos in here, AI-generated or not. And as we're talking about avatars, there's two more releases this week that are actually interesting. One of them is Pika Labs coming out with a product that is meant to give AI agents a face so that they can join your Zoom calls and talk to you. It's a bit weird. This thing is very much in beta though. But if you want to try it, they did open source it, but it only works for the Pika API, of course. I think I might actually try this one over the next week. But one that we'll try right now is Runway Characters. All right, I got the updated app. I'm on a free plan, by the way. And yep, I'm gonna talk to the cat.

Ah, you're here.

I was in the middle. What's up, Mochi?

What do you want to know about me?

Okay, tell me your deepest and darkest secret.

A secret, you say? The deepest and darkest? Fine. I once batted that glass off the counter just to watch it shatter. The sound was quite satisfying. Don't tell anyone though.

Now. All right, Mochi, what else can you do for me?

I can do all sorts of things. I could tell you a story, critique your—

I don't know, man. What am I supposed to say? It's a talking cat with a female voice. You try it for free though. It's kind of fun. Let's see what's next. Okay, the next we have a quick look at SeedDance 2.0. This one went viral when it came out. Video generation model. It was really almost impossible to access. Now it's available on major platforms. And we went ahead and looked at some examples. And run our typical benchmarking prompts. I always enjoy looking at these. Let's start with the sloth on a floaty. Okay, that's really good with the camera movement and everything. This one is always tricky for it because there's a lot of touching water. Oh my God, so much camera movement, but extremely good. What about the ghost? Yeah, also accurate. I mean, it's just clearly visible from these benchmarks why this got hyped so much. Okay, that hand was weird. Overall, I get it. If you want the highest quality video model, it's probably the one and it also lives at the top of various AI video leaderboards right now. Next up is an interesting one and not something that you might necessarily use immediately, but I do want to talk about it because it got so much attention on Twitter and it's a very interesting idea. It's Andrej Karpathy's LLM Wiki idea. And now there's also tools around it that came out. And here's the base idea. He just tweeted that, hey, I'm basically using LLMs to build personal knowledge bases for various topics of research interest. In other words, he just creates text files surrounding certain topics. And then he stores them and his agent searches over them. It's like a personal wiki. There's a bunch of details we could talk about here. But where this got interesting was people testing this versus traditional knowledge retrieval workflows like RAG. And what some tests, and now this is not super verified, but some tests found that it was 70 times more efficient than the RAG workflows. Now look, quick primer if all of these means nothing to you. RAG was basically the workflow of the past few years to deal with many, many files. And this is a different approach to that workflow. With RAG, you would store it in a vector database. And there was a whole process of putting stuff in there. And then you had to retrieve it. That's what the R stands for in RAG. But now he's just saying, hey, just put it in text files, build your own wiki, and let the agents see all of it. No need to store it away. Now, this has actually been the approach that tools like Claude Code have been using over the past year. Claude Code doesn't use RAG. It just stores a bunch of text files and retrieves over them. But he used this to build his own research library. And people really resonated with this idea. Now, they resonated with it so much that it has 19 million views on Twitter, and that people built out various versions where you can start using it too. All of it is in the super early and super experimental phase. This repository is one of them. But I'm featuring it here not because I want you to start using this, just to show you that these new ideas and approaches are being generated in real time by real users. And actually, for some of the more advanced users here, I just want to briefly show you that I really got into Obsidian recently together with my open claw. I built out this entire homepage for all the education that I do. And it's incredible. It's basically my agent's playground where he uses all of this as context and then creates new files in here. But basically, this is my own wiki on my own life and educational style and all the different patterns that I follow so that my agent can be aware of it too. And we just collaborate in one place. Now, this is different from what Andrej Karpathy is sharing. He's talking about a knowledge base like this to just for research on a certain topic. And this is really operating system for my work. But I thought I'd share both of those with you and we can dive deeper in the future if there's interest. One more big story this week, and this one will be pretty short because it's a state-of-the-art large language model from China called GLM 5.1. And this is a big boy. This one you won't be able to run on any MacBook. Heck, even the big Mac Minis have trouble with this one. It has 754 billion parameters. But on many benchmarks, especially relating to coding, it almost matches Opus 4.6, the current go-to model for many people. And the main thing that they market around this and that has been communicated is that it can run for up to 8 hours on a Gentic task. And mind you, this model is fully open source, fully freely available now. Just need a machine or rented GPUs to run it. Insane, especially following last week's Gamma 4 announcement from Google where they released their own version that is actually small enough to run on some phones. It's kind of the opposite of this. This is the big boy. Gamma 4 is kind of the small version. Both of them open source, way more capable than anything we've seen before at that size. So yeah, just AI moving forward. Most consumers just keep using the web apps. That's where the best models are at. One quick follow-up to Suno 5.5 last week. Team member Daniel actually got so into it this week that I just have to report back that according to him, it's actually way better than what we talked about last week. We made a bunch of AI music and used jingles that you'll from here on out hear at the end of these videos. You guys seem to have really liked last week's one. So keep an ear out for that one. Oh brother, this guy stinks. Okay, that line makes me cringe on the inside, but we're keeping it in. Then we have Microsoft releasing 3 in-house models. This is their first move that directly competes with OpenAI. Not kind of, but like directly. They have a transcription model, a voice model, and an image model called MAI, with MAI representing Microsoft AI. And then finally, we have new cloud connectors. Notoriously unreliable, but now you can also connect to your Microsoft 365 account. If you're using that over Google Docs, Google Sheets, you can connect into your Word documents if they live in the cloud too now. Okay, that's pretty much everything we have for this week. A lot of variety. And as per usual, I would love to hear which one of those stories would you want me to go deeper into? I read the comments every week, so leave one or upvote another one that represents your opinion and we can go in that direction. All right, I hope this was helpful or inspirational. My name is Igor and I'll see you very soon. AI news you can use. Breaking down the stories you actually care about. AI news you can use. Igor's here to help you figure it out.