General Intuition Secures $134M Seed to Revolutionize AI Cognition

0:28Welcome to the AI Chat Podcast. There's some big IPO talk. And of course, we couldn't go throughout the day without talking about Spotify's new AI rule and product that they're going to be launching. We're gonna get into all of that on the podcast today. But before we do, I wanted to mention, if you wanna get access to the top 20 AI models in one place for$20 a month, actually almost over 40 different AI models, check out my startup, AIbox.ai. Imagine ChatGPT, but instead of just ChatGPT powering it, You can use something like chat GPT to do some research. You can use perplexity to go and get you links and you can get Claude to turn all of that into a document with amazing tone.

1:07There's a lot of different use cases and you can also get things like 11 labs for audio embedded into it. A whole bunch of really impressive image generators. We have everything from XAI to Google, to deep seek, to meta, all of that on the platform for$20 a month. So if you want to check it out, it is AI box.ai and I'll leave a link in the description. I hope it saves you a ton of money and makes you a lot more productive when using AI tools. All right, let's get into the podcast today. So General Intuition has just raised$134 million. This is their seed round, and they're doing this to teach agents what is called, quote unquote, spatial reasoning.

1:41And they're doing this using video game clips. Now, the story behind this startup is fascinating, including the fact that apparently last year OpenAI tried to acquire them for$500 million. dollars. This is something that neither of the companies have commented on, but a report came out from the information that covered this last year. And there's a reason why they were offering this much money. So general intuition is a spin out from a company called Metal. So Metal is a platform for uploading and sharing video clips, specifically from video games. And they spun out this new Frontier AI Research Lab because they specifically had a huge data set of video game clips.

2:22They have about 2 billion videos per year uploaded to this platform, and they have 10 million monthly active users that are uploading. What's interesting about this is it's actually bigger than Twitch or YouTube for this data set. So they have a really valuable data set. Open AI wanted to get it. They weren't able to. This is what their CEO had to say. When you play video games, you essentially transfer your perspective, usually through a first-person view of the camera to different environments. That is their CEO, Pim DeWitt, who's talking about that. And I think one thing that's interesting is he says that essentially when gamers are uploading these clips, they get a lot of really extreme cases.

2:57They get clips that are very negative or very positive examples of what's going on in the game. He said, quote, you get this selection bias towards precisely the kind of data you actually want to use for training work. So essentially you're getting a lot of edge cases that are hard to capture in real world filming. And it's almost like, I mean, it's, it's basically like you're getting synthetic data of all of these different edge cases from, you know, billions of clips every year. This is an amazing data set. And the reason why it's so amazing is because it's, it's not just for training, you know, like an AI video generator, but it is what we're calling spatial reasoning.

3:31So it's helping AI models to understand the world around them. So it's not just for, you know, like video, but you can imagine for robotics, This is incredibly useful for a robot to watch and be trained off of this video game data. The robot in the real world understands what it can and can't interact with and what the repercussions of these are. And the hardest part is the edge cases. If the robot's walking on the very edge of a roof and it gets too close, the chances are it might slip off and fall because, you know, it could learn that from a clip in this video game data set that isn't really probably a lot of these clips in the real world.

4:03So this is impressive. They raised$133.7 million led by Koshula Ventures, General Catalyst, and Rain participated in the round, top tier VCs. So this is a really, really phenomenal company. I am very excited to follow along with what they're doing. All right. Let's talk about, speaking of the latest and greatest in video, VO 3.1. This is Google's latest model and they're adding it to their Flow video editor. So not only are they coming out with a new model, but they also are adding it into one of their tools. So this is going to make it better. 3.1 has improved audio. So this is kind of the big thing that they're saying here.

4:42They're like, they have VO3, which was a great video model, but now they're increased. They're making the audio better. They have granular editing controls, better outputs for images to video. And apparently it can create much more realistic clips and it has, it sticks to basically the prompt to give it a lot better. So I think this is really exciting. Um, they're rolling this out in their video editor, which they're calling flow. It's in the Gemini app along with vertex and the Gemini API. Um, it's said that since flow launched in May, users have created more than 275 million videos on the app.

5:16So this is something that a lot of people are using. All right. Speaking of new models coming out, we have to talk about Claude Haiku 4.5. This just came out. And the reason why I'm so excited about this is because essentially this is their latest small model and it is insane. I think it's about three times faster than their previous Claude Sonnet or Claude Haiku 4. So it's three times faster and it is cheaper. So I think five months ago, Claude Sonnet was a state of the art today. This gives you Claude Haiku 4.5 gives you a similar level of coding performance, but it is a third of the cost and it is more than twice as fast.

6:00So the reason why I'm excited about this, it does like they, they, you know, benchmark this on the SWE verified, which is a software engineering benchmark. Um, and this thing is like, basically it's beating the, the, the top models that were coming out five months ago. So it's like insanely fast, insanely cheap. And it's as good as what was out five months ago. Um, the reason why I'm excited about this is because we're going to start seeing this implemented into a lot of, um, products that are offered for free, essentially as things are getting faster and cheaper, they're getting so much faster and so much cheaper, and they're still very usable, then we'll see a lot of software engineering products.

6:35You can imagine all these vibe coders, even things like AI box, where we're creating like a no code AI app builder, we're able to offer, we will be able to offer now more free tier options because we have like much cheaper things for people to test out the platform. And if they want the best model, they can upgrade, but I'm really excited about that. And I think a lot of people are too. All right, now for some bad news in AI. Apple has lost another executive to Meta who just is, Zuckerberg is ruthless at poaching Kay Yang, the Apple executive that was in charge of doing their AI web search. He's going over to Meta.

7:13This is just the latest in a whole bunch of people who have been leaving Apple's AI unit. And I think this puts the company in a bit of jeopardy right now. They have a really highly anticipated Siri revamp, which is scheduled to come out in March. It was already delayed an entire year. Then it's supposed to be coming out in March and it does not feel good when all of the top executives who are in charge of that launch are leaving, right? Like maybe it's a big coincidence. Maybe Meta's dangling a lot of money in front of them, but it feels a lot like they think that this launch is not going to go well.

7:44And so they're jumping ship. And even if that isn't the case, I think it just brings up a great, a really important point for Apple to think about the compensation of their executives. Apple hasn't traditionally done some of the crazy things that Meta has done with compensation. And it feels like if they want to keep up, they're going to have to pay more because they keep getting this brain drain to all of the other places. A few weeks ago, Yang started overseeing the Answers, Knowledge, and Information team at the AKI team, which was going to be improving series functionality by allowing it to get more information from the web so they could better compete with opening eye perplexity and Google.

8:22And now he's gone. And Bloomberg is reporting now, maybe this is all just rumors, but Bloomberg is reporting that Apple's remaining team members expect more AI team members to jump ship in the coming months. Yikes, not a good vibe over at Apple right now. All right. Interesting new AI use case is coming out of Kayak. So Kayak has launched an AI mode for travel questions, searches, and bookings. And if you actually go over to Kayak.ai, not sponsored, but it should be, you can try it out. And essentially, these are the features that Kayak.ai is offering. Now, the first thing I will say before I talk about their features is the fact that Kayak is already embedded into ChatGPT.

8:58If you just go at Kayak inside of ChatGPT, you could basically do all these features. But I think Kayak's like, look, we're building all those capabilities for people to go to ChatGPT to use us. We might as well just use a chat GPT API so people could do it on our website too, if they want to. I also think it's interesting that they did kayak.api instead of just adding this straight to kayak.com. Maybe some people aren't used to it. And, um, this is kind of a good way for the people that want it to be able to have a unique place to find it, but not to push it onto everybody. So here's what you get.

9:25You can, um, ask anything. You can plan everything. You have your whole trip in a smart chat. Um, and yeah, you essentially can talk a little bit about like where you want, like just in natural language about your hotel, where you're going, the dates, everything, and it will just put it all together without you having to click drop downs and manually type it all into their boxes. So sort of interesting. Pinterest has just rolled out controls to let you limit the amount of quote unquote AI slop that you are seeing in your Pinterest feed. Now this is kind of interesting because if you've been following along about 10 months ago, people started getting really mad at Pinterest.

10:00There was a post on Reddit on the Pinterest thread. I'll read it to you and I'll read you a couple of the comments. So needless to say, I understand why Pinterest has added this. This is what the person said. There seems to have been a huge uptick over the last couple months of AI slot being served up. This includes pins that aren't linked to any boards I have, or when I try to browse board specific suggestions, I still get loads of AI. Mostly it seems to be AI woman portraits, even when looking at things like DIY projects. I'm guessing the background, there's some SEO terms added. is there a way to block certain tags that pins have, right?

10:34And they said, I've marked them as not interested, reported the spam and blocked the users. And I just get more of the same from whatever algorithm they're using. Okay, and then you read the comments on this post and you realize just how bad the situation is. So the top comment on the post said, I've tried blocking all the accounts I come across that post AI generated pics, but it hasn't changed anything. So now every time I see an AI post, I comment AI slop or AI. Some of the big accounts, which spam hundreds of AI slot picks a day have started turning their comments off. It's kind of funny. Okay.

11:03It's funny that he's doing that, but it's sad because he's actually shooting himself in the foot. If he doesn't want to see more of these, the next comment basically explains and says, I could be completely wrong, but is there a chance that Pinterest's algorithm would start feeding you more AI if you're commenting on posts? I would do this, but I don't want more AI in my feed. And that's exactly what's going to happen. So he evidently is seeing more AI in his feed because of this and I would just recommend not commenting on it. But in any case, Pinterest seems to have fixed this problem and they actually have a toggle where you can choose generative AI interests you'd like to see in your recommendations and you can turn off what you don't want to see.

11:42So the goal is that you're going to be able to start actually filtering some of this AI out of your feed if you don't want to see it. And I think a lot of people are pretty happy about that. The next story I want to talk about is that Nscale is now looking at an IPO. They've just raised a fresh$14 billion in a deal with Microsoft. This is a cloud company, and they're looking at possibly going public at the end of next year. A company spokesperson didn't comment on where, but it comes as the firm is making this new deal. It is a London-based company, and they're going to be providing technology infrastructure to help artificial intelligence scale.

12:20And I think this is pretty interesting because we're seeing more and more of these infrastructure companies get bigger and bigger contracts with OpenAI or Microsoft or XAI or basically everybody. And they're able to then go and IPO. And you also have publicly traded companies like AMD making deals with OpenAI, where if AMD stock price goes from$120 to$600, OpenAI gets 10 % equity of AMD because they're announcing all of these deals and becoming such a big customer. So it's really interesting. It feels like basically all the infrastructure players are playing the Wall Street game. In this case, in particular with N-Scale, Microsoft is going to buy about 200 ,000 NVIDIA GB300 graphics processing units from N-Scale.

13:05And that includes about 75 ,000 GPUs for projects in Norway and the UK that were previously announced. So that is about$23 billion for the total deal, according to a CNBC calculation that they did on this announcement. So absolutely a huge projects going down and, you know, N-Scale is now going to be able to do an IPO and raise a ton of money because they have these huge deals. All right. The last, uh, story of the day that I want to cover is out of Sony. So Sony music group is, uh, they put out a blog post that I'll be quoting a little bit, but essentially they are partnering with universal music group, Warner music group, Merlin and Believe, they're all going to be partnering with Spotify to develop artist, quote unquote, artist first AI music products.

13:52Now I will inter, I'll give you my interpretation on what I think this actually means, but I'll also read a quote from them. So they said, they said some voices in the tech industry believe copyright should be abolished. We don't music, musician rights matter. Copyright is essential. If the music industry doesn't lead at this moment, AI powered innovation will happen elsewhere without rights, consent, or compensation. Okay. essentially what they're talking about um is i i believe they're going to have the ability for you to find an artist you like and to generate music that sounds like that artist now that artist will then be compensated for your streams just like if it was a spotify stream but essentially they're going to be working with all of the big um all of the big players that have all of the top artists on there and you'll be able to listen to spin-offs of a brand new taylor swift album that maybe never existed and taylor swift will be paid for it the reason why they're saying this i kind of like it because it's like, they're positioning them, Spotify's positioning themselves as a good guy, but it's also like almost backhanded blackmail because they're like, they're like, we believe in, you know, copyright, that it's essential.

14:51But like, if you don't do this deal with us, then people are just going to make like, you know, knockoff Taylor Swift albums. Anyways, they're just not going to be like, you're not, they're not gonna be tied into Spotify and the artists are not gonna be making money, which to be honest is true. There's a lot of platforms where you can make these kind of like knockoff, deep fake, whatever you want to call it, albums that sound like an actual artist. And from like an artist perspective, I can see why they just would be mad about this whole situation. And I'll say what they're, I'll explain what their options are.

15:20As a consumer perspective, Johnny Cash is dead. And that was the number one artist I listened to last year. And I would love a brand new Johnny Cash album. So like, I know people will be mad at me for saying this probably, but like as a consumer, I actually think it would be awesome if I could get more music that sounded like the music I love personally. And yeah, anyways, so that's my opinion. As an artist, would I love it if people were like cloning my voice? And I personally, actually, I wouldn't care, but I'm sure there's some people that would care and I understand where they're coming from.

15:53But yeah, as a consumer, I think this is fantastic. This is what they said about it though. There is choice in participation. They said, quote, we recognize there's a wide range of views on use of generative music tools within the artistic community. Therefore, artists and right holders will choose if and how to participate to ensure the use of AI tools align with the values of the people behind the music. Fair compensation and new revenue. This is the last thing they said about this. They said, we will build products that create wholly new revenue streams for right holders, artists, and songwriters, ensuring they are properly compensated for the use of their work and transparently credited for their contributions.

16:25So they're going to make a whole new revenue stream, AKA, in my opinion, there's going to be like a button you click and it generates AI music light in the style and probably voice of the artists that agree to it. This is going to be a new revenue stream. Like I'm, it's not like I'm going to stop listening to Johnny Cash or I'm just going to listen to all this AI stuff, but like, maybe I want like more music that sounds like a, an artist I like, or a handful of new songs, um, to build out a playlist. Something that's sad is like when you have a playlist you love and you spin up, like you just listen to it too much and you kind of get sick of the songs in there.

16:55It'd be fantastic if you could just spin up like basically a clone of that playlist, but all the songs are new songs from that same artist. I know some people for everything that's controversial, I think that would be amazing. Spotify goes and talks about how this is a great way for artist-fan connections to be made. They have more than 700 million people that already go to Spotify to listen to music every month. And so, I mean, it's basically a huge audience that would be possibly interested in this. So more details to come, but evidently they're working on some products and we're gonna have to see what comes out of Spotify with this.

17:29I think it's gonna be very interesting, to say the least, to see what happens with music in the coming future. All right. That's it for the show today, everyone. Thank you so much for tuning in. As always, make sure to leave a rating and review wherever you get your podcast on Spotify. It's the about tab and on Apple music, you can drop a comment. I'd love to hear from you. If you disagree with something I said in the show, please give me a five-star review, but you know, just leave your disagreeing comment there and I will get to it. I assure you. Just kidding. Love you all. Also make sure to check out AIbox.ai If you want to try all of the top AI models on one platform for$20 a month, my goal is to save you money from having to get subscriptions to 10 different platforms.

18:09So if that is something that's interesting to you, lots of new models and saving money, go check out AIbox.ai. All right. Thanks so much, everyone. And I will catch you in the next episode.