#9 Meta Announces Dream Team, Agents and OS's, Figma IPO, Cursor 1.2

Right. Should we try this intro? I'm going to try it without read. I have to read the the headlines, but I'm going to try and read it without I'm going to try and do it without reading and I'm going to look you dead in the eyes here into your soul and tell you. Welcome to Command AI, where we press command F on the internet's weekly news and somehow still can't find our sanity. My name is Samuel Gregory and I'm Kverza and we are here to decode the weekly chaos in AI web design development. This week we have Meta announces its dream team and we are about to start charging AI bots to crawl our content maybe and we have Figma IPO.

Warm Wind, the new AI operating system, looks very interesting. And AI agents for your OS. Uh, and cursor take the lead as the number one coding IDE. There, I said it. The number one coding a IDE. That is insane. But we're about to get into all that on this week's episode. Cool. I only I only up once. I had to look to remind myself where I This was like my f first try. So if uh if you think I can go go again. You were so good. You were so Well, you want to go again? I I I don't know if but I think in the edit it's not too bad that I'm looking on the side because it's just AI your eyes to make sure you're looking at the screen.

So let us know in the chat if it's okay. Two questions, right? whether just seeing the side of my face here because I've got to look at I've got to engage with Kabaza but like here's me looking at the camera. I kind of want to like have a setup where I can just stare at the camera and put Kabaza there so I can like whatever whether this is okay and also whether you think this is too jarring whether you think I'm untrustworthy. Am I an untrustworthy character here like this? That was just my comment. Don't take it personal, but if I see you with these glasses, I'm like, "This guy is trying to scam me or something."

I mean, I am, but you know, right? I'm going to I'm going to tweet about this because I unintentionally jumped straight into um I unintentionally jumped straight into that intro. Where am I going? Let's try Let's try see if we can bring some people over from LinkedIn. Um do actually, did you want to start on the the the meta stuff or shall I start on the meta stuff? No, you can you can start on that because I wasn't sure is this about the meme that I shared which we'll we'll talk about. We'll talk about the meme, but no, this is like this is serious talk. So, the actual dream team that we've been chatting about has finally been announced, I guess.

And it was in combination of a leaked memo from Mr. Zuck himself, but also a tweet from Mr. Alexander Wang. We'll get into him in a second, but I mean, it's the who's who of AI. Those uh nine figure signing bonuses, which I've come to learn are actually not lower than the actual salary, I think. Um I think that's how I understood it anyway. But Alexander Wang, do you know who Alexander Wang is? No. Uh, so he he was the world's youngest billionaire at the grand old age of 25. He became a billionaire and he was the founder of a company called Scale AI. Now that has its own set of drama because do you know what Scale AI is?

Never heard of them. Okay. So basically they crawl the internet. In fact, I've got the website up here. You can see my screen, right? Yeah, they they they have all of the AI data and resell it to all of the I just went into the Matrix. Wait, you're back and I noticed that you're back from from the stream. Hey, welcome back. We have some I've seen some things. I went into the future and then went back and then came back to the present and I've I saw some horrendous things, but I think we'll have to leave that for another day. Sucked into the matrix there a little bit. So, I was just saying that I know this website from their design.

That's kind of up. I don't know what the company does. I have no idea. I just know these discs, these designs, they they were a trend uh and people were talking about them. Oop, sorry. Don't didn't realize that would just overtake you, but um so I guess I guess Streamyard cut me off here, but uh Scale AI, they have all the data. They have all the data all these companies use for their AI. you know that instead of this is this is in some ways how people can or companies can say they haven't scraped your data because they have not scraped your data scale AI have and they've bought that data from scale AI and they're bound to whatever scale AI tells them but saying all that Alexander Wang uh were invented them and founded them and uh became a billionaire at the age of 25 24 I think it was crazy do we know how scale AI works like how how do they well they scrape data they scrape data and and store it in vectors and you know all that kind of stuff standard standard procedure stuff I guess but I mean I can't say that with absolute certainty but it's just it's just data they just hold the data and and uh store it in a way that can be consumed by AIS I'm sure it's way more complicated yeah I I understand that what I'm trying to get into is uh you heard about Um I I think it was a case about Claude uh about Anthropic where they bought books like physical books.

They paid for them and then they took off the cover and they they took off each page and scanned it and then fed it into AI and it was a lawsuit and apparently they won the lawsuit because they technically they bought the books, they used the content as of like as a fair use that judge uh recognized it because hey they are not literally copying you know the content to give it to you. they are just interpreting the text and that's what the AI does but they are probably going to lose the next case uh in which they just downloaded a bunch of books. Well, let's get into that in one of the latter uh articles that we'll be getting into.

But uh because I think it's relevant but continuing on with this who's who of the dream team. Um well the other guy who's leading it of note is GitHub founder or CEO rather Nat Freriedman as well. So it's interesting how he was CEO of GitHub and now he's being brought in as an AI person. Maybe he's just a really good CEO. I don't know. But in this uh in this memo, Zuck announced or said that that Alex was a founder of our generation or something to that effect. So he rates his leadership style well maybe not leadership. Founder is different from leadership. You can found a company but it doesn't necessarily mean you're a good leader.

But um he does praise him specifically. So yeah, CEO of GitHub is in there as well. As I say, uh, looking through here, pioneered RL on on chain of thought. Uh, chain of thought is a a new way of um, accessing information um, and processing information. creator of the O series models of OpenAI, co-creative GPT40 voice mode, uh co-creator of GT40 image generation, built 0304, you know, all these people are heavy hitters and I reckon they're causing a lot of jealousy, a lot of turmoil within Meta. But bringing it back to Meta's onslaught to super intelligence, they actually given themselves a name. They call them the Meta Super Intelligence Labs or MSL and they are taking no prisoners.

And we heard Sam Alolman say that's not how you build a culture and all this sort of stuff. A as Rhymar said last week, winner takes all. And yeah, and Zuck knows that and he is throwing everything he can at this group of people in order to get super intelligence. At the end of the day, nicities, I don't know. I'm sound like a corporate now, aren I? But, you know, nicities can only go so far. So, I don't know. I don't know what to make of this, you know. Um, if we could draw parallels to football, Chelsea did the same thing. Chelsea Football Club, they just went off and just bought the a millionaire um bought the club and just bought up all of the amazing players.

And you know what? They won. They won the Premier League and they they showed everyone well didn't show everyone anything, but like they showed that you can just buy the best people, put them together. These are professionals. They know how to work. They know they've got mouths to feed. I mean, you know, they could feed a whole nation with the amount they're getting paid, but uh morals are being thrown all over the place. You know what I'm saying? Yeah. And I I find this actually to be very interesting. Zach shifted from throwing everything, including the name, at this metaverse thing. And now it's all turning into AI. I'm wondering if they are going to change the name to something with AI.

Probably not. Meta is I think meta is still a cool name though. I don't whether I know it's with the metaverse but um I mean it's a meta name. It's it's it's meta. It's it can it can always stay relevant because it is literally meta in of itself self-descript. We are getting into philosophy here about what Mark Zuckerberg does. But yeah so talent is obviously one side of the thing. Um we we know that they are we know that they now have this they have unlimited amount of money and they have a lot of data. They don't have as much data as Google has but they have a different type of data which is also very interesting.

It would be interesting to know how this work out for them building their AI because I bet they are going to scrape every Instagram description that you've ever wrote all the way to every Facebook post and I don't wonder if they scrape some of your messages that are not end to end uh encrypted. Well, it is worth saying here that you're talking about traditional AI. You're you're probably talking about Llama, the big one, the big boy. Yeah. Now, the what they're trying to do here is yes, they're trying to make Llama a heavy hitter, but they are trying to reach it's in the name trying to reach super intelligence. Yeah. So, that's where the race is.

And whether that's scraping, like I I think we have all of the data. I just And I don't know, you know, whatever. Screw at me in the comments. I think we have all the data we need for super intelligence. We just need the methods and infrastructure to be able to create super intelligence. So I don't think this necessarily creates more scraping and more stealing of data. I don't know. I think this is this is taking the data we we've realistically already have and and re-engineering it in such a way just like DeepS did. They they showed us that you can um you can you can interpret data and run an AI on like half the budget of these other companies.

Interesting take. Interesting take because uh I'm talking to a few AI friends and they actually believe that we don't have enough data and they don't think that we have a a correct um or the right infrastructure and also architecture to reach super intelligence. We don't know like there is no direct indication that shows if we continue this path we get into super intelligence because right now we are just uh putting all the bets on essentially language. We are trying to get to if we are talking about super intelligence we are talking about also consciousness and if we are talking like a super mega conscious super intelligence being uh putting all the bet on language to get there not sure and uh some of the people that I'm talking to they are also not sure so let's see so so I guess my comment I I agree with your with your friends I think my comment is more about scraping more data.

I don't think I don't think we need to scrape more. We need more data, but I don't think we necessarily need to scrape it, especially when you've got Alpha Evolve. Uh not Alpha Evolve, sorry. Uh zero, whatever, where these AIs can rewrite their own foundational their own foundation model in things that they've learned so they can self-improve their own um uh model. I think that we need to give it more data for sure. We just don't necessarily need to scrape the internet for it. That makes sense. Yeah, we need data from other areas. We need we need to yeah give it more data. And at the end of the day, a lot of these models, they claim not to use your data for their own internal models and stuff like that.

So again, the I don't know I don't know enough about this to actually make too much of a comment, but yeah, I I was just responding to your comment on like scraping literally scraping data. I don't know. I don't know whether that's part of this role, but I ain't I ain't getting paid hundred billion or 100 million dollars signing on bonus. I obviously don't have the expertise to even comment on. Not yet. Nearly there. I'm probably about I don't know one millionth of the way there. 100 millionth of the way there. Um anyway, let's move on. What time is it? Yeah, let's move on because we were rudely interrupted by Streamyard. So, similar to what we were just talking about about stealing data, about scraping, Cloudflare announced content independence day.

And in a nutshell, what this essentially means is that the default should be that AI agents or AI is blocked from scraping your data. Um, most of this article gets into the history of Google, which is basically that it seemed like a pretty cool exchange between Google and you. If you've got a bit of content that you want to get out there, a blog or something like that, Google will index your site and in return and people will find you on Google and they'll come to your website and happy days. There's a bit of a mutual thing there. I mean, over time it's sort of swayed a little bit with adverts and people gamifying the system and um all that kind of stuff, but he does go on to say now that we should by default block agents because they are inevitably taking the traffic away from the content that they're now using and there should be a exchange of money between the content that they are taking it from which I haven't had an awful lot of time to digest this concept.

cept but I do kind of like it. I like the fact that if we're giving if they're you know it's just a simple exchange of of services. I'll give you data if you give me money. And I think that this could grow into something that might mean that you know we they took us by surprise. You know we chatbt comes out and all of a sudden everyone realized oh what for the last few years you've just been learning from our YouTube videos. We the it was a sneak attack right. Yeah. Um, and so this is what they're, you know, proposing. And I think what's happening is that, um, Cloudflare are going to be doing this.

He didn't say anything about the the actual implementation of this, but they does specifically say that they want to they're just like, they're just calling it content independence day. um along with other of the world's leading publishers and AI companies to default block AI crawers unless they pay uh crawers unless they pay creators for content. This is where they come in and say, "Well, I'm just doing this for free." They ain't doing this off their own back. They're going to aim to build a marketplace. We'll work on a marketplace where content creators and AI companies, large and small, can come together. So, they're they're obviously going to try and monetize this thing. Um thoughts? This is very interesting.

First question, these AI crawlers, what are they trying to do with my content? Because it's one of these two things. One is scraping for teaching the LLMs. That's what you mentioned by learning from our YouTube videos, which by the way are really difficult to make and it's really valuable information if you think about it. It's very especially my videos. They're super valuable. Like so much value. Buy our course. Watch. Well, it's very domain specific, you know. It's it's not something like that you can randomly just come across or think about, right? when you make a course about I don't know react like your react course it's not just that you sit there and think about it and then it just comes to your mind and you can output it you have to try things you learn a bunch of different things so it's very valuable so that's one thing crawling your data to teach these LLMs so selling your information for that doesn't sound bad the other thing is just for not scraping it but crawling it to show it show it as an answer to potential let's say buyers or consumers.

You write a blog post you you basically put all your learnings from like two decades into that blog post. We have such blog post you know um and chat GPT you know they they take it from your website and they they just show a small like you know number or any of uh any other chat app they just show you know that small link indicator that I bet 99% of people never click. So I think that's also fair to be paid. Um if you are let letting if if you are serving my content to consumers and you are charging them to serve them my content, it's just fair to say, "Hey, give me my share."

Yeah. Yeah. Interestingly though, and I do like this, but I'm just I'm thinking of Dia right now, which I have Dia, mate. You uh you Okay. I'm I'm I'm I'm transitioning. I'm transitioning. Yeah. Um it in because this is what I'm talking about with training of data, right? Like DR in theory doesn't need to train on your data. Okay. Well, I mean will and whatever, but it just needs to know where to fetch the transcript from what service or whatever or just go and fetch that transcript from the from the API or whatever. It just needs to know how to process your request and process the data that it can get. It doesn't need to train, you know, and arguably is it it's not really preventing anyone from they're on your page, right?

They're on your they're already on your video, right? It's it's not um I is DIA using any API because from my understanding it's just looking at the HTML just the content because if it's on a YouTube video you see the transcript of that YouTube video on the page it's already being loaded so it's tapping into that uh same for emails so I but hold on that thought because I think DIA inevitably will go that direction to become more agentic. That's that's just the next step. Like okay, right now you're I'm going into my email and asking you to write a reply. Next step would be I don't even go to my Gmail. I just ask you to reply to that email that I saw on my notification on my phone.

And you just and Dia just opens the tab, goes there. Like theoretically, I I bet they already have a prototype doing some sort of like agentic stuff. Well, I hate to park the conversation because I think we could pick this up on another article we're about to go into. So, stay tuned on this one. Put a pin in this because I've been thinking about this. Yeah. I've been thinking about like you asked me like three or four weeks ago like where is all this going and there are there are multiple way directions it's kind of going but to keep it like on track with the cloudfare thing and paying and you know compensating people for the data and stuff like that.

I don't know. I think I think on the surface it seems fair like it does seem fair that they're training on on data but like I think like I someone said from anthropic I think even if we completely cease development on AI right now it has the potential to change society. So I think we're kind of already not in a bad way, but just like I I don't think we you know we can is it con if for a convenience thing. Uh someone's asking if we're live. Yes, we're live. We can see your comment. We'll get back. We'll get to it in a second. Um thanks for tuning in. What's it? What's your name? Uh Edan Nidic, if I said that right.

Um, yeah. So, it's it it's an interesting proposal and the fact that they're going to do some sort of marketplace, I'm not too sure. And then how how my content on YouTube, how are YouTube going to take a cut of that then? Platform fees or YouTube going to take that, you know, have I signed an agreement with YouTube that actually they take any kind of scraping or whatever that I've already signed my life away with that. It's it's interesting though where this where this stops or whatever. Um but yeah uh we have had a comment in which we should address and I'm curious to know uh if I show this. Oh nice. We can just show a comment.

That's pretty cool. Have you seen Miyagi agents in web flow? Have you seen Miyagi agents in web flow? No. So yes I have seen Miyagi agents. I am waiting for it to mature a little bit but it is news and it is interesting and uh it's basically vibe coding for web flow and uh what's it um Bradley Huber I think and XAM I think are teaming up on it if I Bradford Bradley Bradford um yeah they built like an agentic um layer in front of the web flow CMS which is actually something I was really interested in doing. Um, but it's hard to find the motivation to do such a thing. I think it's pretty straightforward.

I I would I would describe it would you describe it as an agentic uh sorry, not an agentic an MCP like a more complicated MCP layer for for Web Flow, but seems interesting. Um, waiting for it to mature a little bit. They've not released it yet. Are you looking I see your face wincing. I feel a little bit bad uh for this app. I to be very clear I have no internal information about like if web flow is going to you know release a vibe coding or something like that tool but I bet they will and their version will have more you know more access or will they just buy out this thing it's like why even get started when we can buy you know no disrespect probably a couple of indie developers are probably a bit cheaper just to buy their tool I bet they and flash 100k in front of their face, they'll be like, "Yeah, we'll take that."

You know, probably uh I mean, probably. But but we've seen this again and again in uh iOS and in Android world. This is not something new. As as these platforms grow, the community starts developing for them. And these platforms grow so big that they take some of these very essential ideas and they develop it as an internal integrated tool. And then these apps that one day actually power the platform, they go dead. They just die out because people just use the internal tool. Yeah. And it's more trustworthy, more convenient. And this there in lies the issue with relying on some platform and we all are we're all in some way relying on platforms in some way shape or form.

You know I built products and whatever superbase goes down I'm I'm If uh if some API changes I'm I'm not building everything from scratch. I do think we're going to get to a world potentially where we aren't necessarily relying on external platforms. But you're right. The point is is that they can see this. They can easily, it seems like a logical direction to, you know, we were talking about it um way back when Webflow AI first even announced. So, this isn't like rocket science, but it seems cool, seems interesting. Um, Eden says it's going to save it seems like it saves a lot of time. I damn well bet it does save time. I agree.

So, yeah. Uh, Web Flow. Was it Miyagi for Web Flow? So, it was naturally going to come. You are asking for it, like I say, but it does look cool. Yeah. Um, cool. Let's let's jump to a quick break and we will be right back. Support for this episode comes from Flowbase. If you are building in Web Flow, Frame or Figma, Flowbase can help you build much faster. With over 4,000 components to choose from, you have a huge variety of wireframes and super clean, nicely designed sections that you can put together by just simply copy pasting them to your project. Over 15,000 icons as well, over a,000 illustrations. They also have a super helpful Web Flow app called Boosters.

With boosters, you can add things like sliders and marquees and countubs to your project with a few clicks and without any coding. So yeah, big thanks to Flowbase for supporting the show. If you want to try them out, go head to flowbase.co. And just for you guys, I want to show discount command flowbase all capitals for 20% off any plan. This is a limited offer, so get in there early. All right, that's it. Let's continue with the show. We are killing it with segways in this episode. I have to tell you gone from web flow talking about web flow into our semi-web flow advert. So, but we are about to take a bit of a U-turn.

Um you probably and this is where we get into start get into the conversation of this where is AI going and and you know staying inside of DIA. Uh warp. I will not stop talking about warp. I really want to make a video on warp. It's it is a terminal um and it can run it basically picks up whether you're just asking the AI a question but it will just it will use that to navigate your system. It's a terminal. It's a terminal but you can use um AI prompts to do stuff. Great thing about this I can say copy this file to this location or move this file or install this package whatever through natural language and it will just because it has terminal access it will do it.

This is not about Warp as a product. I just thought I'd drop that in there. But they have introduced AI agents. And the way I'm seeing warp or way I'm using warp right now is as a a an I'm using it like an operating system because it have has access to my files because it has access to applications because it has access to running because most whether you know this or not the computer the the applications that run on your computer they all are able to be accessed and commanded with through the terminal. you know, I can't speak to to to what extent they can they run. I'm going to stick on a fan cuz I'm bloody boiling.

Um, they, you know, so I can I can through natural language say create a folder here, open this app, do this, do that, and if I forget a command, I can just ask the AI to do it. Now, they've introduced agents, which I think adds another layer of like my I'm basically Mac OS has Apple intelligence as far as I'm concerned. you with warp like I I can ask it simple questions like as you would any sort of AI. I can choose my I can choose my um maybe we can you can it's got an auto button here and stuff like that. So, as far as I'm concerned, I'm feeling like Mac OS has Apple intelligence through Warp.

Now, I I it's a referral code that I'll put in. But genuinely, I really really like it and I think agents are just going to unlock that completely even further and it's a really exciting thing. I don't Does this interest you? like the the whole terminal thing is this do you think this is a okay so terminal is not for me yet because I'm still like getting used to I know getting into code uh more and more but it's not about code this is the thing it's natural language to say open this app and you know when I see these things uh typically I'm trying to you know think think of use cases um I'm I'm not any like the use case for this, but I see I definitely see the use case or like the war that all these tools are trying to win the on the OS level to be to to be the agent that you want to be.

I know this word agent is now becoming the new meta but previously we we used to say assistant. Uh all of these are becoming that assistant that we all wanted and dreamed of. Yeah. Yeah. Um and Dia is the closest to me. So this is why this is why this is all um questionable. It's why it's all up in the air. Oh, sugar. Uh warm wind. Slightly weird name. It makes me think um of things. Warm wind OS. Now this guy, he does this in one take. Fair play to him. English is not his natural language. And look at this. Look, this video just looks beautiful. Looks like real AI generated. No, real. I don't think so.

It looks really really good because the whole video and I don't want to get copyright struck, but it's like you can see the shadows here. They're moving around. He's just real. Yeah, I think it is real. And to get to the point, whirlwind or warm wind OS is this thing and and this starts to remind me of deer. So I think you can give it a bunch of apps that it has access to and you can through natural language. If we turn these subtitles off right now, wrong one. You give it the prompt and it can do things within these applications. And it's not just again it's not just uh using the APIs or it's not using MCPS.

It is literally like the rabbit R1 promised. Move. See that mouse moving? Yeah. It's copying. It's copying from Gmail. Gosh damn it. My pointerism is it's copying from Gmail. Emailing a bunch of people transferring data over to spread Excel spreadsheets and then contacting them on oh no sorry it's contacting people on LinkedIn it's basically running a mail out automation and it's all done through this prompt and what he demonstrates really interesting so it's a layer abstracted further from Dia because Leah Dia aims to tackle this idea that everyone is in the browser and they are but I've got OBS open right now I've got Final Cut open I've got mostly browser I got notion open all this sort of stuff that Final Cut isn't about to be in the browser you know for professionals we are professionals you know we we we still use applications and this is where this is demonstrating this old um real estate app which then this is used by hundreds of thousands of people across Germany they're not about to update they're not to add an MCP server to this beautiful piece of uh trash here but the AI agent within Warmwind can understand and navigate this application.

So, does this mean that I can stick in an edit in Final Cut and say just edit this video in Final Cut or set up a stream in OBS or something like this and or run an automation just to I should probably I could probably do this in Raycast to be fair. One button click and open Streamyard. It opens YouTube Studio, opens OBS. Yeah. And then we've got our live setup going on. All that to be said, where do you where do you sit knowing that Oh no, one final thing. He is using it on an MacBook and I'm wondering obviously Mac famously don't sell or lease out their software. Is this a application? It's probably an application that's running on Mac OS that can that has access to you know or or as a virtual machine that can run other apps within it.

I don't know I don't know what the architecture is but this is I think technically uh these apps can do this right. So how how does it work when you give access through like any desk you know I know it's on Windows but uh screen sharing screen sharing yeah well not not just screen sharing but giving access so the other person can literally control the the the mouse the keyboard enter data so they can fully interact with your computer I think even Zoom has this so I believe like this as a as a piece of like as the infrastructural piece exists. I believe this is an accessibility option that uh these u have. Yeah. So technically I don't think that's the biggest issue and I understand this is like a a level deeper actually uh more you know has more access than DIA.

This is the the the device that could have access to DIA as well including other browsers as well. Right. But I think the hardest thing that these uh companies are trying to figure out is the AI models that need to get good enough that they can reliably do these tasks. right now they can't and also that they are cheap enough because I I bet these are like burning a forest for each video that you want to edit because they have to take a screenshot of your app for every single frame if it's a video editing uh task, right? Well, I mean, yeah, but I guess it once it's taken those screenshots, it could potentially do anything with that.

But I the your point is taken. I have I have something to say about the the So that would be that would be like interesting to see how it's going to be tackled. I'm a bit skeptical because honestly I don't see a lot of um progress with a bunch of these AI tools. I see that we went from ChatGpt 3.5 to whatever chat GPD that we have now. In terms of text, we have close to a miracle. It's a miracle. It's amazing, right? But anything beyond text, I'm not convinced that it's, you know, that level of revolutionarily quality. It's still below average. And for these bigger context tasks, I see it's falling apart. when when you say go do this like like holistically do this task I don't I I I don't have any examples of any AI that can do that big task even in dia the whole premise and the point is that you keep the context and it's kind of like a you know small window of context but that's the point it's very neatly separated and I bet if you would put all the context of all my 200 tabs open into DIA the system would just they would forget about everything but somehow our human brains that that's fascinating to me that we keep that context right like 500 taps you still like you go back to it and you know ah okay I know where I left I just pick it up so let's see how that progresses yeah I think that's that's where I was going to what I was going to say it's like I I think we we need to look at where this is going.

It might not be it might be the first, but it might not be the best. Like these these are just trailblazers who introduce a new concept, a new idea that someone else could stand on their shoulders and be like, huh, they're doing something quite good. Maybe Apple are watching this and thinking, you know what, we could probably build that. That sounds probably easier than personal context, to be honest. Um, but yeah, I think the the it's the opportunity that it's that's uh creating. It's the it's the excitement I think with all that. Yes, you're right. You know, the the skill the the ability right now probably isn't there. It's not we're not saying it's there.

We're saying that look what can be done. And there is a live demo. You know, he apparently it's completely live. If I can um show this. Oh, we should do this one, shouldn't we? Um, and it's kind of sped up. All of this is just it's it's a recruit. Here we go. Recruiting workflow, you know. Hopefully we won't get copyright struck on this. But is this not similar what Microsoft showed? Uh, it was from Microsoft, right? The the the one that could remember everything that you are doing, but I'm not sure if it was agentic. Yeah. So, that is I think you're talking about recall. Yes. which takes screenshots of your and you can just and all it does is process those images and tell you what's on them which is pretty really all it is is taking screenshots of your screen.

It can then it just every AI agent can take a yeah um uh an image and interpret it. So yeah um it's cool. It's interesting. Yeah, we have this same though. Yeah, I don't think it's the same but I bet Microsoft can do the same. So we have this comment uh I always confused what the is agent. Everyone has different definition. I'm not sure if everyone has defi a different definition but do you want to give like your definition what what do you mean when you say agentic? Agentic is different. AI agent a agent AI agents I think it's uh an autonom autonomous like an I I'm going to try and say it without saying the word agent, but it's an autonomous thing that's going off and doing something in the background.

Uh you can have multiple agents doing multiple different things that all come together to accomplish a task. So you're sending an AI off to go do something autonomously and it will come back and do it. I'd be interested to know what Well, first of all, Kabaza, yeah, go on. So for me um when we say AI agent it's a okay let's actually let's say it a different way the the way I see it is you can do two type of things with an AI you can ask it or you can tell it to do it so you typically go to chat you chat to it and you say what is this how to do this and it tells you how to do it but the next step is the next obvious step is to tell it to do it for you, not just to explain it.

Hey, how can I find that button on in my or like how can I change that setting on my Mac? Why do you want AI to tell you how to do it? Why not the AI just because if it knows it should just click it for you and go do it. And the same goes with filling a Google sheet or doing your work for you, answering your emails, uh buying bread for you. I don't know. Like anything is theoretically anything is possible. And um yeah, that that's perfect. Leave it there. Yeah, that's it. I think that's it. Why the difference between asking and doing? I think that's a really nice elegant way to put it.

So, how does that answer your question, Eden? Let us know. Um, we skipped around because I I thought Warmwind was very relevant to Warp agents. Do you want to chat about Figma? Yeah. So, Figma is Figma is apparently filing for their IPO. IPO is when you have a company and you want to go public, meaning that people can buy and sell your shares. Uh, when you start a company, you you start it private. you sometimes don't ask for money from anybody that is bootstrapping. I'm now explaining the entire ways of like different ways of building a company. Another diff another way is you build a company, you ask for money from investors. They give you money with the hope that one day the company IPOs.

That means that company goes to public and public buys the shares and these early investors they can sell their shares and get their money back. And this is what exactly Figma is trying to do. And all of this makes sense to keep it within the context of or like to talk about what happened to Figma one and a half two years ago when Adobe the you know beloved Adobe that we all love and praise wanted to love it. They wanted to buy Figma for 20 billion dollars. That's crazy. uh they attempted it but it failed because the regulatory didn't the regulatories around the world I believe they didn't allow them or I think they they made it hard so they dropped it they dropped the case and it was a big hit from what I understand to Figma because they they've been working hard and they hope that they can sell to Adobe and now they would have quite a lot of monies imagine all these people working at Figma They've been working and getting salaries obviously but these people they've been getting also shares in internally right so at this point they could sell it but didn't happen and what happened instead Adobe gave them a billion dollars because it was a part of the contract so they could Figma at this time what they could do to figure out how they can go for an IPO which they are doing.

So now they are filing all these documentations. So showing that they are uh very I I think they are profitable right they they have a huge amount of revenue. I'm not sure actually if they are profitable but they have uh 228.2 million in quarterly revenue and net income of 44.9 million. I think they're profitable. So that makes them to be they are a bit over $900 million per year. So that is close to a billion. They are close to making a billion. Um and they are IPOing for 15 to 20 billion. Right? That's that's what I heard. I'm not sure what's in this article, but yeah, that's also a good number from what I understand.

A company that is making a billion a year IPOing for 20 billion is not bad at all. That shows that everything is healthy. But let's see if they hit that. Okay. 12 billion. That that is that is a good number. Mhm. That is a good number. Are you going to buy their shares when they IPO? And by the way, this is not an we we are not uh what is it called? Like this is not a financial financial advice advice. Do not uh buy their shares with the money you don't have. So I just went to dinner with my friend Al. Shout out Al. He always watches my stuff. So give us a comment here.

Thank you Al for supporting the show and watching. We we work together quite often and he told me today that he was asked to build a UI, you know, asked to and he's not a designer or anything like that. He he he owns um motion motion graphics and 3D company, but for what he wanted, he was able to produce perfect UI for um on I think I don't know some AI platform. Chachi PT. I think he was actually using Chachi PT which was able to give him a good UI. I think this is only going to exemplify the fact that designers and developers and this isn't unique to Divi designers. there is going to be a lot less need for them or there's going to be a lot less of them because you can get so much out of one designer nowadays if they're using midjourney to create this or if they're prevising a bunch of stuff.

If you learn to previs in midjourney and you're creating draft after draft to finalize a design, you're no longer creating that in Figma. Figma may end up becoming a less valuable and less relied upon company because you're spending less time in Figma, you know, uh, and less designers are around. Saying all that, maybe I will chuck some money over to Figma. I've been using them for for a while. I think we're still a long way off of that. Remember, when I say these things, I'm talking about years in the future. Years and years in the future, maybe, you know. So maybe I will maybe I'll chuck him a few pennies. On my end, uh I'm a paying customer.

I'm paying for myself and my entire team. Um it's my second biggest uh cost in business. Probably probably soon to be uh the biggest cost in business in terms of software after Web Flow. There are like neck and neck right now. Uh Figma and Webflow. And I'm I know it's expensive, but at the same time, it's a software I I really enjoy spending time in. And if they could figure out how to do AI within the canvas, I'd happily continue using Figma. And this is where we don't know how Figma is going. And that's why the IPO, I mean, it's always like this. You are never sure how it's going to be. But they are ipo IPOing because right now they are in a good position and not just Figma there is no company that is sure about their future like look at if you look at Figma okay you ask yourself what happens with AI and design you look at web flow what's happen what is going to happen with AI and development we don't know you look at lovable you wonder is it going to infinitely grow is it going to you get bigger or like more fe featurerich than web flow will it become so so much more convenient uh and good that I use it instead of web flow so we don't know any of these that that's the honest answer that we don't know so your bets are as good as I I would say anyone anyone's at this point and only time can tell if Figma can win but I I'm going to bet that they have a good chance because they have a great team.

They've been dominating this space. Like they have over 90% adoption. That is insane. Like n over 90% of like companies uh especially like the companies who are in this space they use Figma. That is insane. Uh and I bet that they are making you know they are acquiring other companies and they they are trying to figure out this AI thing. So if AI becomes a thing because still we we talked about everything being you know still prototyping but if it becomes like actually reliable and a thing then Figma will be in a good position because they're investing in it and if it ends up not becoming a thing Figma is already in a good position.

So I don't think anyway I would not bet against them and I'm happy to say this because I I I really like Figma so I'm also a little bit biased. But I mean at the end of the day, I mean, you still haven't answered the question. Would you pay a few pennies? I I will. I will. You will. Okay, fine. Yeah. Um, the end of the day, mate, you're a business. It's not just about liking Figma. As Rhymar once said, they are these businesses, they aren't your friends, right? Yeah. They make you money. They make me money. Yeah. I pay them because they make me money. the moment they stop making me money, I stop paying them.

They look elsewhere. And yeah, and that's why Yeah. So, it doesn't make smart business sense to uh to to to leave them right now, you know, because you're making more money than they're costing you. Um Yeah, exactly. Well, I'm either way excited for the to for the IPO. This is my first IPO. Why am I saying my IPO? It's not mine. But it's like the first company that I'm, you know, so closely involved with. I mean, I'm much more involved with web flow and, you know, even with the team. But even Figma, I feel like really close to it because I've been using them for so many years and kind of like feels like I mean that's that's probably not a good sign when it comes to financial sense because you kind of need to be a bit you know objective.

Yeah. Objective, emotionally detached, all that kind of nonsense. So yeah, but you know, we're not financial advisers. We don't know what we're doing. We chuck money at companies that we think are doing good and we think there is opportunity there. So I think there's still a long huh we vibe code and vibe invest we vibe invest our way through life baby and I do think FGMA probably got a bit of a you know a way ahead of them especially if they like you say keep adopting the AI stuff and they use it strategically and we use it strategically as well like I think it's silly to like think that AI can just you can just vibe design a new UI but I think you can vibe design a button and you I think you can vibe design again these prototypes, these previses to get the client to a point where okay, now we're at a point where we can go in and actually nail this design line by line hand designing and all that kind of stuff.

So, um yeah, so we are on to our final Are we on our final one? Yes, we are. Well, you haven't said your meme yet. I've just scrolled down and seen your meme. What is that? Is should we wait? You I I don't know. Um but do you know that the the left that the guy on the left? Yes. Hang on. We can't we can't talk about this on screen now. Yeah. But right now teasing. No, no, no. Go go go with uh we talked about warm wind. Now now we're talking about Yeah. Cursor. So cursor. Well, actually this is this is not a nice tweet. So, they've actually updated their pricing. I have not seen their pricing update, but apparently they've gone from unlimited to limited pricing.

Well, actually, let's just say the news. The news they've updated it to 1.2. Don't know where 1.1 went, but they've updated it to 1.2, which introduces tasks. Um, where is my little notes here? Uh, it come on now. Um, actually, I've just got tasks written down. Maybe I just failed to to uh finish that one. But in fact, let's go here. Let's talk about um they they they've improved the long horizon task as well, which means it can basically run for a long long time. You can search pull requests. You can cue messages up. So, while it's like doing its thing, you can like line some stuff up. I hope that all gets sent as one prompt because you can save a lot of tokens by making sure that everything's in one prompt.

I tend to just um draft out while it's doing it and then and then press enter when I'm done. But sometimes you can accidentally press, you know, enter, whatever. So queuing messages is quite nice. Uh and so 1.2 is pretty good. Now the the to-do list is something I've been really playing around with. I did some vibe out lives last week, this week, last week, where I was just like just checking some stuff out. And one thing I've been trying to nail is this thing called Taskmaster. So if we go to taskmaster.dev I think it's called taskmaster is a way to um create a prd product requirements document and uh yeah I don't know PBS living twinning I don't know something's going on there um a way of queuing up uh creating tasks for your product that you then get the AI agent to run through those tasks and code it all up and all the rest of it.

So taskmaster is gaining a lot of traction. This methodology of creating tasks for your for your coding is gaining a lot of traction. And so cursor now are claiming well they've introduced tasks natively inside the UI more let left a bit sorry we say something can can we see the UI for tasks? I I think it like freeze the frame because is this not um is this not what you showed me? Well, you you said that you are making PR of the in cursor anyway. So is this just not a list a a bulleted you know is not is this not just a list of uh text and what is so special about it except so I think UI and you know being a nice UX as well.

So I think it's a combination of two things. It's it's reasoning. it's showing you the reasoning and it's more so that it's it's planning out and giving you like the opportunity to step in and and whatever to make sure that it's not going to go off and do you know just some random crap for you. So, I don't know if this creates a file, but what I often do, and have I got warp even open, but what I often do is I ask my AI to create a plan and then save that plan to a markdown file for it to refer back to later on. So, that way I can go and work on some other stuff or whatever.

Probably should utilize background agents to do this and and all the rest of it. So I to answer your question, I don't know what the UI is on this one, but I do think it's it's just the way to now AI assist coding, right? It's it's creating tasks. It's breaking those down into subtasks and Taskmaster enabled that, but now we're getting it natively. I'm sure you can break down these tasks into further subtasks to be honest. So I think that it's um a welcome addition and it's just something that more people need to do. I've got a video recorded just talking about like my top tips with core code and how we should be coding and stuff like that.

So I think that we are starting to amalgamate, is that the right word? We're starting to just finalize what AI assisted coding looks like. And it's a combination of memories. It's a combination of tasks. It's a combination of planning. And it's it's it's putting less emphasis on the actual writing and understanding of code and more in the planning and the architecture around the the lines. I think I think that's where we're going. So yeah, it's a like I say, it's it's a welcome feature, but I don't I don't know whether this creates a a file, but I do like having those files to be fair. And I probably would tell the agent to export this to a file um and create a bunch of tasks like someone said something, but apparently I think it's basically butchered version of what Einstein said.

If he's given a problem, he's got an hour to solve a problem. He he he'll plan it for the first 50 minutes and actually execute in the last 10. I think that's where we need to be. That's kind of what they've always said how we need to code. You should do more planning, whiteboard sessions and stuff like that, but I think ultimately we just get lazy and we just want to just dive in and just make the change. But I think we just need to step back a little bit and just spend more time thinking and enjoy that process of thinking and planning and whatever. We're too we're too used to just needing things like at the just in time.

Yeah. But but uh to to get into the fundamental of this uh we say writing it down and planning all of this is to understand that's the the fundamental issue that we are trying here to solve to understand what the scope of the project is what the tasks are and what um each piece of our software supposed to do. So that understanding is what we've been aiming for. Even when we were in school, I remember like learning mathematics, the teachers would say reading and understanding the question is 90% of the way into solving the the problem. I'm rubbish at that. Low key, I'm rubbish at that. And it's and it's literally the same thing. And guess what?

Even if you get into the writing and making and vibe coding of the software or like anything that you are creating, even if it's like a physical thing, even that process of making is still learning and understanding how the pieces work together. You're exploring, but you're exploring in a different way. I and I wouldn't even go and say it's a bad way. I I think it's actually probably the better way because that's prototyping. That's understanding. And guess what? The second time you are doing it, it's it's much easier. Why? Because you understand how pieces come together. So, back to your point, the thing that you said about Einstein or I don't know who else uh it was.

It's the same thing when you understand it fundamentally the problem that you are trying to solve, you can solve it much easier. You know how to systematically approach it. And the second time you write a piece of software, it's order of magnitude easier because you know how to pull it together. But arguably, you know, we're getting to a point where we're not solving the problem that we are we are becoming the I call them shepherds. I've always said we're the shepherds. We are becoming the lead devs and the agent is our we're hiring the agent to do the work for us. So but the better direction we can give the agent, the better direction we can give our young spunky employee cla code or you know coding thing of choice we are becoming managers you know and I think that's the understanding needs to come from the architecture and how it's put together and as you as you dig in yeah you're going to need to know what files do and and and you know the nuances of this stuff, but we don't necessarily need to know how to solve the problem.

We just need to know how to uh communicate the problem to be solved in in more effective ways. So, I see what you're saying, but AI definitely flips the head on actually doing work right now and actually where where the doing of the work sits. Sorry, we we are still doing the work, but what we are doing is just moving a little bit. Yeah. So and that is exactly that is moving and like a analogy that I give to that is we are moving from being the pilot having a co-pilot to us becoming the co-pilot having AI being the pilot. So right now we are doing the thinking, we are doing the planning and the AI is helping us and now it's moving some knobs and you know turning some stuff and it can click a few more places.

It's our co-pilot and and you know not I'm not talking about like the Microsoft copilot just in general you are in control. Yeah, you're in control and the co-pilot is doing these things and we are moving to that direction that we will lean a little bit more back. We don't know how far that will be, but the the AI becoming more of the pilot and you know saying okay I'm in charge of the plan and I'm in charge of like the execution and you're just there to make sure that you enjoy the ride. Maybe. Do you know what I thought you were going to say? I thought you were gonna say we were beginning to be become the plane.

I really want to become the plane. You are the You've been the plane all along. Yeah. Now that's meta. That's uh We are the plane. Um there's some TV show that's creepy as hell where they all these It's like Thomas the Tank Engine but for planes. It's really really creepy. Um anyway, that's that's for weekend watching just for you guys. Don't know the name of it. Yeah. But I think the kicker, just to wrap this baby up, the kicker has come that cursor have now reintroduced a limited pricing. So it used to be unlimited um completions on sorry, sorry, unlimited prompts or whatever on their 20 bucks a month plan. They've now, you know, they've moved say floor onto the API pricing onto the onto the max plan.

Um, I'm I'm This is what This is how these companies get you. I honestly don't know how to interpret that from this. But maybe go back. But why is no one talking about the cursor making the pricing unlimited then changing it to limited without warning? Check the dashboard and they charge me $71 for just one pl one day on Sonic 4 usage. So, be cautious. Oh, and they've they've um not allowing API custom a uh your own APIs, which is actually a bit rubbish because I think you were able to hack that and I use the word hack very lightly here, but you were able to use open router or some AI agent through the custom API solution.

Obviously, that seems like it's buggered now, but people are getting 500 errors. So, it's it's bit of um it's a bit of a shame really, but I can understand why cursor did it because I think someone responded somewhere where they were like people were just running up like huge bills in just one session and they just literally just could not could not cover those costs. And it brings into the question I want to find the subsequent tweet that I found maybe it's in these chats here where someone was just explaining that you know having this token based pricing is confusing af to try and follow and understand and it's unpredictable. We try and simplify it by saying hey we're just going to cover the cost and we're just going to say per message.

That's basically what they did here. But it just people are just hammering it and it's they're just not able to to to cover the costs quite simply. I'm still looking for this tweet that basically explained that, but I'm not able to find it unfortunately. I'll probably flush it up on screen if I can find it later on. But we're in a sticky situation where either anthropic needs to become cheaper or well, I don't know. I think Anttop just needs to become cheaper really. It's just so bloody expensive. It's very interesting like even with the current pricing uh from what I understand all of these AI companies are losing money, right? Well, yeah. I mean here's this is now you raising a really important point.

Here's the worry. Here's the worry that they're all losing money. That that is very interesting because where is this going? like okay it's getting cheaper like the smaller models but we still see a huge amount of like context needed. We need a lot more tokens and you know we are still nowhere close to you know making big big jumps. So, it's very interesting. Uh, and it's a bummer because I I like cursor. Are we going to keep our plan like the cheap plan that we got? I freaking hope so. I freaking hope so, but I would pay for it anyway. Actually, I don't know cuz, you know, this is off topic completely, but I've been thinking more and more, and we started briefly talking about at the beginning of the episode, more and more about running local models and having a machine powerful enough to run local models.

I think I Oh, we were talking about like how many forests are we burning down by running all this stuff and all this like I think the answer will be that people will become well no I don't think the answer this that's just me being hyperbolic with my statements. I think that there will be a large chunk of people that running your local because it's really easy. I've got a video out on running a local LLM. I'll have another video running uh coming out on running a vibe coding on a local LLM. Fortunately, my machine just isn't powerful enough. But I would like to I would like to have a setup where I can just run local thing.

I don't have to rely on a server. So, I have a 1490 Ti here laying around. Just try it. Just give it a go. Give it a go. Um, that's a PC, right? You That's a PC. It is a PC. And you know, my I think you just need a llama. I I think you just need a llama or an open web like my my my power is a thousand watt and my electricity bill is com you know comparing the size of my apartment and everything I do is pretty low mainly because I use I don't use that big ass PC. I sometime at night I game with it, but it's like a short amount of time and mainly I use my MacBook and it's like just using 30 watts.

That thing is going to eat at least 300 watt at its base. Um, so have you seen that have you seen that PewDiePie uh video where he's I have it on my watch list. Yeah, it's really good. Really interesting. It's really inspired me to think if this doofus can like run everything. Yeah. Yeah. Yeah. Sorry. Go on. Um, running everything off Raspberry Pi. And I wonder if there's a world where you can run your local AI off of Raspberry Pi. Obviously, you need your GPU attached to it, but I'm talking about basically removing any kind of unnecessary like I I don't think an LLM needs CPU. So, it's just all in the GPU. So maybe it's um Does it need a screen?

Does it need Sorry, a V. It needs VRAMm as well. A lot of Isn't that in the GPU though? Yeah. Yeah, it is. But it needs a lot of it. And like Mac is in a good position because you know like have you seen those Mac mini farms? No. that it's like a room with a a thousand Mac minis in it because they are a really good value for the price because you get 16 gigab bytes of like RAM which is the same as the VRAM uh for the GPU. So like the LLMs can tap into that. Um anyway, what were you gonna say about PewDiePie? He's becoming a full-on engineer. It seems like that's and I I think that's really cool.

But what what um what worries about me or like what doesn't work for me with um trying to running my models um locally is I I hate to say it but I like to maximize. I I'm not a minimalist. So if you give me, you know, if you give me a model that runs on my PC and you say this does 98% of what you want to do with it, it's powerful enough for that and in return you wait just a second for the answer. Or you can have this model from OpenAI that costs some money but it does you know a little bit more. I picked the one from Open AI. You're the one burning down the rain forest.

The best one I want to burn down all the Yeah, you're you've got your You're the one in the cinema with their phone on max frigin brightness. 120 Hz screen all the time. Yep. Yep. I I I want to maximize. I'm not saying that you have to take make any sacrifices. I just I'm imagining a world where people will be more inclined to run local LLMs, you know, and I think we might get to a point you've got a break you've got a GPU. You got a freaking I don't know what you know what GPU is in 1490 Ti. It's like if you're a maximalist, it's got to be the best, right? It it's well, it's the second best because it's from last year, but or like one and a half year ago.

Yeah. Yeah. They they now have the 5090 which are like huge and expensive and not like not that much better. But yeah, um it's I'm worried that the quality won't be the same and I'm worried that if I do this, I see another model that is significantly uh more capable and my PC can't handle it. So I have one of two choices. I have to either go back to these platforms that burn forests to use their model or I have to burn the forest in a different way by upgrading my PC. And I have a friend, a developer friend, he has, you take my graphics card, I believe he has six of them now. He's building a rack at home literally.

And he has trouble like figuring out the the electricity for it. you know, these these things eat up power. And one potential interesting story that we can uh maybe sometime cover I we need to go and learn about it, but I'm like genuinely interested in it. It's not news. It's about how these companies handle these really big like infrastructure infrastructure tasks because when you I've I've learned about a little bit about XAI and Elon Musk, they are they are going nuts with creating these um basically warehouses to house all of these GPUs and there are tons of like engineering um challenges that they have to overcome. in order to you know to run them with the right amount of electricity to get the electricity in the first place because the grid is not ready for it.

So in term for X AI I I've heard that they are burning uh fossil fuel at the factories at the warehouses like they have a bunch of generators because that's the fastest way you can do it. And I've heard even that other companies are doing similar things because the grid cannot handle this amount of energy. So they are burning literally like gasoline at the at that place that they are located in. Um because also the land is cheaper there, right? So they have all of this and they buy a bunch of carbon credit so that they can offset the carbon. It's it's like crazy the the things that are going on with uh on that regard.

It's it's like Bitcoin mining all over again, isn't it? Yeah, it is. It is. And it's easy um for us to to just say, well, these are very expensive. And if we look at it, yes, it is expensive for a reason. And you see, holy it could be even more expensive. And there are like a bunch of more work to be done. So that makes me uh kind of like hope that even though AI is like on one end trying to come for some jobs there just we humans figure out infinite amount of jobs to create somehow. I don't know but let's see about that but yeah a lot of engineering challenges it starts to put me in a really dark place when I start to think about all this stuff.

So uh I think we should stop there. Yeah. So speaking of engineering and talent, I want to share this meme. I don't know if you have well uh I see your screen. Can you see your screen? There we go. Now I see my screen. So don't know if you've seen this. On the left we have So forget about becoming the next Ronaldo. The new Meta is becoming AI engineers. So apparently Meta has paid $100 million to this guy and just to join them and who who knows how much he's getting paid a year. He's getting paid uh quite a bit more than Cristiano Ronaldo. I mean, I do not I cannot overstate how incredibly incredibly if you don't follow football, Cristiano Ronaldo is so famous.

Like, football is just a monster of a sport. Like if you're in America, which a lot of my viewers are in America, I don't think you can comprehend football as as a as a sport in the rest of the world. Ronaldo is just absolute just I don't know how would you describe him? Like he's just he's just so famous, ridiculously famous. And for that to be, you know, for this guy to be paid 20 million on top of that is something it's insane. He is literally the most followed person on Instagram with 653 million followers. That's insane. That's insane. That's more follow subscribers than Mr. Beast has. Yeah. Yeah. He has more than half a billion followers on Instagram and he has I think a few billion followers across all his social media.

I think it's over a billion. I don't know if it's a few billions, but I know it's over a billion, right? And here's the thing, right? Arguably, Cristiano Ronaldo does not contribute a huge amount to society to warrant an $80 million signing bonus, right, to Man United. arguably it's entertainment and the argument is like yeah and also you need to remember that imagine you were working for a company and you knew you were making them hundreds and hundreds of billion dollars and you were getting paid a salary of 20 20,000 or whatever it is right but you knew you were bringing to that company hundreds and hundreds of mill you want a piece of that pie this is my argument for why footballers get paid a lot of money they are they would You would be livid if you knew you were bringing in so much money and you were just getting paid a tiny I bet Ronaldo is getting paid a tiny chunk of that money that uh that the both FIFA and Man United bring in.

It's pocket change probably. So fair play. Uh who is the guy Yan Yahoo? I don't know how to explain that. I probably is probably going to be contributing a lot more to humanity than Ronaldo ever would. Yeah, probably. Yeah, especially if you It depends on how you value things. But I I I think we agree on on this uh premise and I I think he brings a lot more and it he's getting a lot paid a lot and it's um it's very interesting because Meta is obviously they are making a ton of money and they're the people who are working at Meta they are they are not getting paid uh too bad especially the ones located in the US or North America they're getting paid pretty well, but a lot of people at Meta, they're not getting paid really that handsomely.

But this guy, because he is apparently really good, I don't know anything about him. I just know he's a AI researcher, um, Meta is trying to hire him from, you know, like take him away from Open AI. So, they give him a 100 million signing bonus. I believe it is in uh form of some of it is cash, some of it is uh shares in the company and things like stocks like that. That is very interesting. And this is again yet another week and another reference to Silicon Valley the show. They literally do this. They liter especially nowadays with age of AI. So, uh, like a slight, um, kind of like spoiler, it's not exactly a spoiler because you don't know the show, but one of the companies hires people and doesn't give them tasks.

And these people, they just go on the roof of that, you know, office in the company and they spend all day doing literally nothing. They are actually dying out of boredom. And these are like people that get hired just to stick it to the competition. Oh, really? That's the only AI researcher. Like it sounds so like what what are you doing? Like can't you get AI to do your own job? You're just doing some mathematics. Uh well, in this case, this guy is obviously going to do something. It's not just to stick it to the competition, but I bet that's a part of the deal as well. Probably. Probably. Let's wrap it up there. Yeah.

Good episode. I've noticed we're both wearing white as well. We're becoming one. Yeah, becoming one. Well, we we we might already have become one for uh non-English speakers if they are listening to us. That's true. That's true. Um, hello. Hello. I'll feed us in. Um, it's really messing with the with the dubbing. Cool. Well, I've I've got I've got uh some drinks tonight with a friend and then tomorrow I'm going to see Oasis. So, that's going to be my weekend. Yeah. Cool. Awesome. Got an exciting one. Cool. Cheers for tuning in everyone. As always, like, subscribe, buy my course. I don't have a course. Oh, I do actually. That's actually I do have a course.

Yes, whatever. Good one. Yeah. Jokes, jokes, jokes. Keep on vibing. Peace out.