Bankless - AI ROLLUP: ChatGPT's Viral Explosion | xAI Acquires X | Coreweave IPO Rocks Markets

Episode Date: April 3, 2025

AI heats up, markets shake, and the internet gets Ghbilified. OpenAI hits 1M users in 60 minutes, CoreWeave IPO rocks NASDAQ, and XAI acquires X. A new “world’s best” model quietly drops—did Google just take the lead? Cursor raises at a $9.6B valuation as “vibecoding” enters the mainstream. Welcome back to another week of the AI Rollup. ------ 📣CRYPTO TAX CALCULATOR | TAXES MADE EASY https://bankless.cc/Cryptotaxcalculator ------ 🪙FRAX | SELF SUFFICIENT DeFi https://bankless.cc/Frax 🦄UNISWAP | SWAP ON UNICHAIN https://bankless.cc/unichain 🛞MANTLE | MODULAR LAYER 2 NETWORK https://bankless.cc/Mantle 🌐SELF | PROVE YOUR SELF https://bankless.cc/Self 🏦INFINEX | THE CRYPTO-EVERYTHING APP https://bankless.cc/Infinex ------ TIMESTAMPS & RESOURCES 00:00:00 The Ghibli Effect https://x.com/alifarhat79/status/1904884015858884884 https://x.com/unusual_whales/status/1906831621166440583 00:05:09 Coreweave IPO https://www.ft.com/content/519006a9-82c6-4939-afa1-58921a399105 https://www.cnbc.com/2025/04/01/coreweave-shares-bounce-ipo-price.html https://x.com/TheStalwart/status/1907438208662552647 00:14:28 xAI Acquires X https://x.com/elonmusk/status/1905731750275510312?s=46 00:20:02 Grok's Telegram Integration 00:24:38 New Best Model Has Arrived https://x.com/bookwormengr/status/1905050740110090429?s=46 https://x.com/gfodor/status/1904980153329242147?s=46 00:32:09 Cursor And The Vibe Coding Explosion https://x.com/ArfurRock/status/1906768733135098360 00:39:02 AI Crypto Trend Update https://x.com/defi0xjeff/status/1904154090424402163?s=46 00:45:47 BillyBets March Madness https://x.com/playprophetx/status/1899535745435742608?s=46 https://x.com/bunchubets/status/1904224002123473294?s=46 00:56:51 Olas https://olas.network/ 01:01:32 Vana https://x.com/anna_kazlauskas/status/1905414468315549737?s=46 ------ Not financial or tax advice. See our investment disclosures here: https://www.bankless.com/disclosures

Welcome Bankless Nation to the AI roll-up where we stay up to speed with all the emerging trends, developments, drama in the AI crypto space. I'm David Hoffman here with my co-host at Jaws. At Jaws. How's your week, my man? What's up, dude? You know that meme of Dave Chappelle kind of like scratching his neck? And he's like, you got any more of that stuff, David?

Yeah, yeah. That was me for the last two weeks because we've got a ton of AI updates to kind of go through on this episode. We skipped an episode last week. Sorry for all those. the regular consumers of the AI roll up. Just schedules didn't work, but we're back with a double dose of AI. Yeah, and there is a lot to go through. Some really important stuff happened over the last week, David.

For example, the most contentious and polaric events ever were giblified and animated to be made super cute. We all got a new word into our like vernacular, which is giblification. Yeah. And it refers to a studio jibli style of artistic animation. which took over Twitter and the world, essentially, because the folks at OpenAI basically integrated this new feature, which allows anyone to upload an image and, you know, giblify it per se. So it made the world pretty cute.

It gives generalized artistic direction to what they want that photo to be, like, amended to. And then everyone was like, make it studio gibbley. I mean, David, like, pull up this threat that I just sent you. I mean, we kick off with basically Trump after his... The Trump fist pump, yeah. Yeah, the first pop after his like assassination attempt. And I don't know, for some reason, it's, it's kind of even more epic than what was originally. It is more epic.

You scroll down and you got the Tiananmen Square. I saw a studio Ghibli version of like the World Trade Center, like in the midst of the plane going into it. I'm like, wow, I am surprised that they are allowing this to come out. This is incredible. It's pretty nuts. But, okay, so like a more under-talked point around this. whole givably. Okay, it's really cool. It took over the internet. Everyone had a load of fun. But two important things happen kind of behind the scenes from like a business and economic standpoint for Open AI, David.

Number one, Sam put out this tweet where he explains that when they had that initial chat GPT moment, right? Which was around 26 months ago, I think he says in this tweet, it took them five days to get to a million users, which at the time was record breaking. You haven't seen that across any other company. When they did this animation image generation feature, which led to the giblification of everything, they added 1 million users in 60 minutes. That's over one hour. And that was two days after the feature launched. So I bet you that it was...

One million users in five days gets collapsed down to 1 million users in one hour. It's a pretty insane acceleration. And it's kind of like a trend that we're just seeing with the AI ficat. of everything in general, right? But then there's even more news here, which is they recently closed their round, conveniently enough after this new feature launch, which now values OpenAI at $300 billion, David, with a B.

So it's pretty insane to kind of like see their valuation growth from, I think it was like $10 billion to start off with, and then they've got like a new capital raise of $40 billion. That's $40 billion. I thought that was an April Fool's joke until I realized it was posted on March 31st. So it is the largest private tech funding round in history per CNBC. That's pretty crazy. That's pretty crazy.

Yeah, I think one big theme of the episode this week is there is actually very much a lot of life in the AI markets. Despite like the decreasing NASDAQ, decreasing SPY, like macro fears, the AI economy, the AI investment, the AI private markets are extremely. strongly strong. The fundamentals aren't tracking the market share or the market cap growth of a lot of these different stocks and tokens as a result of that. All right. So that's the first bit of the news that we're going to cover this week. The rest of let me kind of preview some of the subjects we're going to talk about. We're going to talk about the core weave IPO that just rocked NASDAQ. And again, gives us a sense of the strength of the AI bull market. XAI acquires X. So the AI company has acquired X, formerly known as Twitter, so we'll talk about that. There's a new world's best model

out there. There's a new world's best model every single week, but this is a new week, and so there's a new world's best model. So I want listeners to place your bets, which AI lab created it? Did it come from the United States? Did it come from China? Did it come from a different country? Everyone placed your bets. And then the company behind vibe coding is now valued at $9.6, almost $10 billion, as vibe coding works its way into public culture. At least, on Twitter. So these are all the big news. And then we're going to, of course, get into crypto AI. We're going to talk about Billy Betts and some things going on in the AI agents and crypto AI side of things.

Ajaz, are you ready for Corweave or any last comments before we just dive into it? No, let's get into it, David. The OG Crypto AI company well before anyone realized that this was an intersection. They were useful like mining and stuff, right, David? Coreweave, yes. Okay, so this Corweave company, which just went public on the NASDAQ just this last week. Started off as a crypto mining company doing Ethereum GPU mining back in Ethereum's proof of workday.

So it was founded in 2017 as Atlantic crypto. Like I said, initially focused on cryptocurrency mining with GPUs. In the 2018 crypto downturn, the crash of the ICO meta, Corewe pivoted to leverage its GPUs for just high performance computing tasks, not necessarily AI, a lot of like rendering stuff. like gaming stuff gaming rendering like graphics rendering yeah just like maybe like maybe Hollywood studio CGI kind of rendering things like this because AI still really as an industry hadn't really taken off yet the demand for compute resources were still still pretty nascent so this company is based in

New Jersey where there's just a lot of data centers but the core we've also run data centers all across the United States and Europe with Nvidia GPUs just to deliver scalable high performance cloud computing for AI models and inference that's their business model today just to talk about some of the numbers. Corweave reported $1.9 billion in revenue in 2024. Net loss of $900 million just to investments, just trying to scale up their business model. During the IPO, they raised $1.5 billion, making it the largest AI-related listing at the time. So that's kind of the big headline news is the Corweave IPO, $1.5 billion raise made it the largest AI-related listing ever. So that's pretty cool.

So everyone is kind of looking at the IPO of CoreWeave to see like, okay, markets are down. There's like shaky markets. Trump is like making it real hard for equity investors. And also we're all kind of looking at Nvidia being 30% off the highs. We're like, all right, like is their life in this AI investment like meta or is it coming to an end? Like give us like the canary and the coal mine for AI related investments. And so the stock opened at $39, which is below the IPO price at $40. So it opened up below $40.

It kind of had this lackluster reception on day one. And then a bunch of people, commentators, we're talking about the obvious short around Cori for a few reasons, very high debt load, $8 billion of debt and $2.6 billion in operating lease liabilities. And so this raises concerns about the company's ability to just meet their financial obligations. And also there's customer concentration. So 77% of its revenue of CoreWeave's revenue comes from Microsoft and OpenAI, Microsoft relating for 62% alone. There's some business model doubts that people are wondering if CoreWeave is just, you know, GPUs in the cloud. How defensible is that moat? The critics have argued that the lack of unique

moat to protect against this market position. It just doesn't exist. And then also just the market typing and around AI hype. Like I said, the IPO came around a volatile market period with tech shocks under pressure. And some people are just saying CorWeave is just overhyped because AI is overhyped. It's trying to ride the AI wave without actually any proven profitability or diversified revenue. And so this was the day one of the IPO. And so the headlines here in the Financial Times says CoreWeave's IPO tested the waters and missed at the mark. That is day one. So that's a lot of fun. Day one, a lot of fun. A lot of fun. A lot of fun. Can you. It came in like $1 below the IPO price,

IPO price of like $40, maybe $40 and a half dollars, came in at $39. That's day one. Day two? Not the same story. Rips, 42% higher, way above the IPO price. We have Joe Wisenthal, who tweeted out this morning, Correve up another 20%.

And you look at the price right now from the bottom of the market where it bottomed out at $36, like $4. below the IPO price. It got all the way up to $65. It's currently at $56 this morning. That's a crypto chart in a bull market. That is a crypto chart, bro. This thing has ripped on the second day after the IPO price.

So any IPO purchase, anyone who has bought Coralweave in the last like two days is well into the green. Wow. And I would say that this is, if we are looking at just like a canary in the coal mine for broadly AI, looking real good. Looking real good. I got more to say about Corrieve, but I'll pause and let you chime in here. Well, no, I mean, my next natural question is why?

All the FUD made sense, David. Surely the FUD like tracks into price progression similarly to how all the AI progress is leading to price decline, right? Well, what happened here? What am I missing? I don't know. I don't know. I think you can maybe use Corrieve as a litmus test for inference,

demand for inference. Okay. And I think this is coming, the timing of this as well as the giblification of the internet, I think it's pretty good because Sam Altman was tweeting, you guys are melting our GPUs. Right. Like we need to actually take away this feature away from free accounts because our GPUs are just melting. I don't think we have any data as to the costs that the giblification of the internet has put on OpenAI just for inference costs. but you can just see Jevin's paradox just exhibited in real time last week

where there is so much demand for turning one picture into a different picture just to share it with your friends, right? There's just a lot of inference demand and this is what Corrieve says. This is one feature. Yeah.

Wow. Okay, so there's another part of this story that I think is pretty cool. Okay, so if the Ethereum, here's a fun fact. If the Ethereum merge happened on time, Corweave would not be here today. Explain yourself. What?

Okay, so this is where CoreWev started. They started off mining Ethereum in the proof of work days, and that was their revenue. Oh, they would have never focused. They were, they, from the early days, they had apparently just understood that in the future, there will be a lot of inference costs, but that future is not yet here today. But that was apparently the vision that they, like, founded the company on is like, oh, yeah, GP. GPU resource demand is going to be high, not just for AI, but also just modeling graphics and also cryptocurrency. And so their plan was to use Ethereum GPU mining to just be sufficiently profitable to get to the point where, like, in the future, AI inference will be a thing that can actually float their business. And so that was back in 2018, 2019, they just started buying up a ton of high-end Nvidia GPUs just to build out a deep well of capacity.

and because the Ethereum merge just got delayed from 2019 to 2020 to 2021 all the way to 2023, they were just able to maintain profitability until ChatGPT got released in 2022. And that's when this AI inference like bull market really, really happened. And so I think that is one of the biggest, most like they saw it, they saw the vision for AI inference and demand for AI inference. They also saw that, you know, Ethereum mining and GPs. mining can carry them to that point. And I think with also a ton of luck, they were able to just pull that transition off pretty

beautifully. So I think it's a pretty cool story. Yeah, that's awesome. It seems like the right pieces fell together at the right time. That is insane. Like I wonder what Kowee would have ended up. It would have just focused on crypto mining, presumably, at that point. Yeah, they probably just would have shut down and be like, okay, we just made our money.

There's, once the Ethereum turned into proof of stake, like the next largest proof of work, like yield was Ethereum Classic. which was like one hundredth the value of Ethereum. So once Ethereum proof of work went away, there was nothing left for them, except JATGBT had already launched. I'm curious, David, over the last 24 hours, how much have they added to their market cap? Do we know?

Just in the last 24 hours? Well, this is a live chart, right? Yeah. And so it is the second. And so, I mean, the market. Well, like, even from its bottom, from like, its IPO price. Like, it must be up like... So the bottom on March 31st was 36.

$36. So at the very bottom, at $36. and $15. To where it's right now. And its top was like 65? It's up 56% right now. The top was a plus 80%. And what is that in terms of like market cap?

The current market cap, the peak of the today, the current market cap is $26 billion. I think it's down to like $22 billion. Wow. So it added like $15 billion, if not over like the last $1,000. couple of days. Yeah, I think it added like eight or 10 billion dollars since for the bottom, since the bottom. Wow. Wow. That is crazy. Anyways, congrats to Correif.

Yeah, and congrats to the IPO holders and stock purchases that just ignored the entire foot. Yeah, that's crazy. All right, take us into XAI, uh, uh, uh, uh, uh, okay. So big news from,

uh, the Elon Musk, uh, X or Twitter sphere right now. So XAI is essentially the offshoot or subsidiary that we initially thought, that built GROC. GROC is basically the Twitter AI model. And GROC, we've spoken about GROC, GROC 3, its latest model, outperformed a lot of the leading models when it launched. But big news this week is that XAI, so that's the company that built GROC, acquired X,

aka formerly known as Twitter. And that is a pretty insane thing to happen in my... opinion, right? So if we pull up this tweet from Elon himself, he starts it off with saying X has acquired X and an all-stock transaction. The combination values X-AI at $80 billion and X at $33 billion. But what I really want to focus, David, is on this fourth paragraph down here where he says, XAI and X's futures are intertwined. Today, we officially take the step to combine the data, models, compute, distribution, and talent. This combination will unlock immense potential by blending XAI's advanced AI capability

and expertise with X's massive reach. And this is the point. Twitter or X currently has 600 million active users, David, right? And we know that, like, Elon is like obsessed with like finding the truth and information or yada, yada, whatever. But the fact is this model, this AI model now probably has one of, of the largest distribution mechanisms or, you know, captive audiences to test any kind of product or feature with them, on them, right? I don't know about you, but like having GROC 3 available on

Twitter has made it super easy for me to like do my research on Twitter or like catch up with the latest news or, you know, check out sources for a particular tweet to see if, you know, what they're saying is real or not, right? And now we've kind of spoken about this before, but we believe that X will become kind of like a founding place for a lot of different AI apps in the future, presuming like that, you know, they allow agents to prosper and AI models to exist. And this is just like confirmation, in my opinion, that this is going to, this is the path that X and Elon and the entire XAI team wants to lead. The third point I'm going to make, David, is the data that this model now officially has access to is going to be insane. And it's in real time.

It's people's opinion, subjective thoughts. It's a truth-seeking machine. All that data is going to be fed into the AI model, and it's going to be ragged, right? Retrieval augmented generation, which is going to make the model smarter and smarter in real time, per day, per hour, per whatever unit of time. And I think that is just an insane access point to have. And it's crazy to see it like officialized, right? I wonder how much of Elon's original motivations for purchasing Twitter actually was.

in this story. I think it's hard to understate how incredibly valuable the information is that comes out of humans Twitter accounts, doing, just reporting about what they see about that end of the world. Like, news happens first on the internet,

on Twitter. And we've seen, I always would talk to Ryan about the difference between reading articles and just being on Twitter specifically. Where like after, but there was like, my life was different, I got on Twitter because previously I would read about news events about things that were happening. And then I got on Twitter and I realized that news reporters are just reading tweets.

Yes. And turning that into articles on MSNBC.com. But you can just skip the like editorial or just the MSNBC or whatever interpretation of real life events and just go straight to the source, which is Twitter. And that is the unique compelling case for XAI, which is, the data that only they have, like, high fidelity access to, which is people's tweets. And so I think the real timiness around the Grock model, I think, is going to be very valuable. And you can see, you can see why, like, Elon Musk is, like, putting these pieces together.

I'm going to be completely honest, you guys. I already thought XAI and X were already intertwined. Yeah. I didn't know that they were not already intertwined. They were intertwined, but XAI was a subsidiary of X. Okay. And now it's become like the owning entity of the social media site. So that flip is very potent and important, right?

It's weird to think that Twitter is owned by a larger entity. Yeah, it's owned by an AI lab right now. It's owned by an AI lab. Which makes sense that I think you could be in a parallel universe, Twitter is still Twitter and it would still be bought by an AI lab. Well, you know what's a kind of like weird a way to think about it, David? You know how like all these social media sites basically have reviews? smart trained algorithms to like serve you up the best content. The algorithm in this case owns

the social media site. Hmm. Explicitly. Yeah. I don't know where this goes next, but I'm excited for it. Well, one of my, one of the more obvious thesis that I think this is heading in is, I think XAI is going to be on a path of really aggressive market penetration, right? Well, in the same week, actually, David, they announced that they're launching integration into Telegram, right? And Telegram has, you know, over, I think, 300 million monthly active users or something ridiculous like that. Now GROC is going to be embedded natively into chats and into the app itself, right? I actually, I wrote about this. If you pull up this tweet, David, I actually argued that like GROC integrating into Telegram is actually going to be really good for agents because now, right, it's one billion MAUs.

have access to real-time AI. But what I think is really cool here is if you have this AI model training off, once again, personalized chat data or imagine it becomes personalized to your group chat, David, right? Where you're talking about different tokens or you've got a question. You could just ping rock. It gives you a quick answer and you can make a decision on what token to buy or what movie to go book and see or et cetera, et cetera, right? And I think that data is just going to constantly feed back into the,

the main mother entity, which is XAI, which makes its model even smarter and applicable to any other application or distribution network that they're using. So it wouldn't surprise me if we see a ton of different apps appear on X natively embedded that might be agents or just features like OpenAI's giblification image generator over the next couple of weeks or even months. That is the main difference of all the AI models out there. GROC lives in the place, places that I go anyways. It's consumer facing. Like if I need to go to chat GPT, I have to go type in chatGBT.com. I never type in GROC. It's in Twitter. It's embedded in Twitter and there's a button for it. And so that's the kind of the unique thing about GROC for me is it's already in the

places that I'm navigating on the internet. Yeah, exactly, exactly. Imagine a world where your day-to-day banking runs on a blockchain. That's exactly what Mantle is building, powered by a $4 billion dollar treasury and poised to become the largest sustainable on-chain financial hub. As part of their 2025 expansion, Mantle is introducing three new core innovation pillars that bridge traditional finance with decentralized technology. First is their enhanced index fund, aiming for $1 billion in AUM by Q1. It provides optimized exposure to Bitcoin, E, Solana, and USC, complete with built-in yield opportunities. Next, Mantle banking promises to revolutionize global value transfer through seamless blockchain-powered banking services, bridging crypto into your daily life.

Mantle X blends AI with Defy to deliver an intelligent user-friendly experience for everyone. And the best part is that this is all in addition to their already launched products like Mantle Network, ME, and FBTC. Ready to step into the future of finance? Follow Mantle on X at Mantle underscore official and join the on-chain revolution today. The Arbitrum portal is your one-stop hub to entering the Ethereum ecosystem. With over 800 apps, Arbitrum offers something for everyone. Dive into the epicenter of Defy, where advanced trading, lending, and staking platforms are

redefining how we interact with money. Explore Arbitrum's rapidly growing gaming hub from immersed role-playing games, fast-paced fantasy MMOs to casual luck battle mobile games. Move assets effortlessly between chains and access the ecosystem with ease via Arbitrum's expansive network of bridges and onrifts. Step into Arbitrum's flourishing NFT and creators space where artists, collectors, and social converge and support your favorite streamers all on chain. Find new and trending apps and learn how to earn rewards across the Arbitrum ecosystem.

with limited time campaigns from your favorite projects. Empower your future with Arbitrum. Visit portal.arbitrum.io to find out what's next on your web-free journey. You may have already heard about Infinex. Infinex has, in my opinion, the nicest cross-chain swap and bridge feature that you will find anywhere. It is called Switch, Swap and Bridge, and we're going to show you what it looks like. First, we're going to log into my Infinex account with a pass key. Now, there's no seed phrases in Infinex.

This is just a one-click setup with biometric pass keys. But in addition to that, my Infinex account is fully non-custodial. So bam, I just logged in. It was two clicks, and I'm already into my Infinex account. So let's go make a switch. I'm going to go switch my USDC that is on base. And I'm going to buy Barra chain, which is a completely different chain. So we're going to switch this.

I'm going to press that button. And then Infinex is going to execute this order, this cross-chain order for me. And now it is done. But actually, I'm not really feeling bearish anymore. So I'm going to go from Berra to Penguins. I'm going to buy Penguin on Solana. So I'm going from the Barra chain to Solana. See, no transaction signing, no gas to worry about.

You just switch across whatever chain that you want with Infinex. That was so easy. Go check out Infinex and try your first switch today. All right, that's a new best model out there that I hinted about this in the intro. But this is the weekly news. There's always a new best model. There's always some new leapfrogging that's going on. Who leapfrogged who this week, Jaws?

Okay, drum roll, please. Google. Google. Google. Google people. Wait. Who could have seen it? Which one is Google?

That's Gemini. That's Gemini. Gemini is Google. Okay, so Gemini is the new world's best model for the week. Exactly. So Gemini 2.5 Pro, as McKay Rigley in this tweet states, is now easily the best model for code. He goes, it's extremely powerful. One million token context is legit.

By the way, one million token context window means basically you can print, copy and paste essays, essays worth of information. into one single prompt and it'll comprehend all of it. It also means that it has... So one million tokens is like one million tokens is roughly one million syllables. And so that'll collapse down to like 600,000 words. Yeah.

700,000 words. And man, like 1,000 words takes you about 10 minutes to read. David, there was a time where it was literally like 500 words max in a prompt. Oh my God. It's just insane. And if people are wondering, you know, why is this? particular characteristic really important. Well, think about the amount of information you could allow the AI model to ingest in a single prompt. And then like couple that with further up

prompts that you like prompt afterwards. It basically means the AI acts smarter in a much more quick and efficient way, which is really important if you want like to have really intuitive, powerful, magical AI experiences, right? If you have to repeatedly tell, you know, this AI model over like a number of prompts like, hey, this is, by the way, like, this is my age and this is my weight and et cetera. It just gets completely unusable. Another really important thing is it also means it has a really large memory context window, David. So it's going to remember the conversation for much longer as well, which means that, you know, one of the things that I'm doing with chat GPT right now is I'm getting it to act as my dietician and personal trainer in the gym, right? So it has my blood marker

indicators. It knows about my ethnicity, my kind of body type, how much I work out, what I'm eating. And it measures all the calories and everything. And it tracks and it can give me a comprehensive report of my progress basically every week. And you can't do that without, you know, a large context window and a large kind of memory capacity. And Gemini 2.5 Pro is basically hitting all these different markers. I think the most important one, though, David, is it's unseated clawed as the best AI coding companion or assistant. So that is kind of a big deal. It's a huge deal. Cloud I thought had a very sticky moat when it can't come to developers. Correct. And basically, so for those who don't know, Claude is kind of like the leading

air model from Anthropic, which is one of the leading AI model producers out there. And for many months now, it has been the best at helping developers code. And in some cases, take over a lot of their coding work. But now Gemini 2.5, which bared no kind of indication that they were going to be that it was going to be good at coding has now just completely leapfrogged, um, clawed. And I think, you know, that for that to happen in such a small window of time is just an insane to see. Um, so really, really cool. The conversation that we're highlighting here on Twitter, I think is actually pretty poignant. Uh, so let me just read it out. Uh, tweet reads, uh, from this one individual, uh, not inside of crypto Twitter. It's kind of hilarious that Google dropped their

biggest upgrade ever. And Gibley core completely hijacked the zeitgeist. example of how Vibes now overrides tech merit in the intention economy. Interesting point. Then the tweet that we're looking at is responding to this and it says, respectfully disagree. It is true. The timeline is full of chat GPT-40 images. However, people who pay with their credit cards for API access are indeed noticing that Gemini Pro 2.5 is way cheaper and way faster than competitors and appears to be a great model. We will continue to have a divergence between consumers and companies. And I think what he means here is that, okay, great, like the giblification of the internet

is definitely capturing attention in zeitgeist. But people paying money are paying money for Gemini Pro 2.5. And so there's kind of like a quality versus quantity conversation here. And I think what we're seeing is chat GPT has taken over mindchair. Like it's the phrase chat GPT in the same way, Google, you know, just Google it, just chat GPT it. But that's like being pointed at as this, like, like kind of low value commodification of AI and not something that we can convince consumers to

pull out their credit cards and actually pay for it here. But everything that you're saying about the strengths of Gemini is convincing people to pull out their credit cards and actually pay for it. That's kind of the conversation that I'm saying. Exactly. So like kind of like another way to think about it is that, you know, attention economy, how you shape a narrative, mind share is pretty much everything, right? And consumer facing products are the way to do that. That being said, I think they are like two sides of the same coin, David, which is B2C, which is where Open AI this week excelled at with the giblification feature. But then B2B is still a very dominant vertical. And that's kind of where Google is playing in with this particular model release.

It's kind of how I think about it. I think something that's really cool as well, David, is we always speak about these models every week. But we need to like have some kind of comparison chart that we can. kind of relate back to. I really like this tweet from Peter Yang where he kind of like summarizes, and I think he's used an AI model to summarize this, which is kind of funny. But you can kind of like see where Gemini 2.5 lines up in terms of like coding and how JetGBT lines up in terms of image generation. And it is just crazy how like, you know, the flips between these models and their leading capacity in particular verticals is changing just every week.

And I want to kind of like track this kind of like every day. One important thing I think is really important to kind of like look is we see some red crosses on this image, David. And it's to show where a lot of these models are kind of lacking in, right? And they just don't have that feature. Yeah. Deep research, for example, voice chat, image generation, video generation, life camera, computer use. And in my opinion, I think these are some of the features that are going to start popping up in the immediate. weeks to come. And I think it would be worth a discussion in the future as to where these

features might be most valuable, right? I for one, use chat GPT's voice feature all the time, right? I've mentioned this on previous episodes. And I think it's going to give us an idea of how these AI models are going to evolve in terms of like consumer applications, David. Looking at this chart, you just kind of got to tip your hat to chat GPT. I know this chart is opinionated, but chat GPT has the gold star, which, There's only one gold star per category, right? It has the majority of the number one positions across all of the features here. So everyday answers, math, reasoning, deep research, voice chat, image generation,

showing really that chat Djibouti is the frontier model, at least when it comes to just like features. David, have you vibe coded anything this week? I'm so incredibly non-technical that even vibe coding feels too technical for me. Dude, you could literally prompt, you could go onto this app called cursor, and you can prompt it like, hey, can you make me like a flappy bird mock-up? And it'll just do it for you, David. It takes seconds, yeah.

I urge you to create a cursor account because I think this is going to be one of the biggest companies to date coming through. And actually, like, recently on this week, the company cursor, which, as I'm saying, is the company behind vibe coding is valued at $10, well, $9.6 billion, but let's call it $10 billion. after they closed a round of $625 million led by A16Z, Thrive Capital. They raised $625 million. That is so much money. Oh, yes. And for people who are asking why, I can show you like a quick kind of like demo.

So if you pull up this tweet that I linked here, David, it's this guy called, I think, Chong Yu. And basically what he's demoing here is he's put in a single prompt. or maybe a few prompts. And he said, basically, can you build me a flappy bird clone using cursor and Claude?

Claude is an AI model we mentioned earlier. That is really good at coding. And it did it, but in 3D, David. So if you watch this... A 3D, Flappy Bird. But it's still a 2D game,

right? The bird is still going from left to right. Correct. But, you know, you have like very unique kind of characteristics where you can kind of like

edit the map in real time. As you can see, he's like moving around the map. You see sunlight, shadows. That kind of stuff you don't really see in like a 2D sphere or in the original Flappy Bird game.

Also, if you notice, there's a lot of like flow chart action where he's like moving a cursor and he's dragging boxes from left to right. That's how easy it's become to code, David, for some of these different apps. Like, you don't need to really understand what the underlying code is going to be. You just type in words in literal English and you can kind of create this feature, which becomes a box. And then you just move the box around and you say,

hey, can you, this feature, which makes the thing, thing I'm a ball jump up. Could you like make it happen with when I press the space bar please? And they're like, yeah, sure. Like, we'll get that done. Okay, it's edited. It's updated for you, right? It kind of reminds me of Andre Carpathy's tweet that he, I think he made about six months ago,

or maybe even a year ago, David, where he goes, English is the new, is the hottest programming language. It's a new hot programming language. Yeah, I really like that tweet. Yeah. Yeah. And it's, it's just kind of maintain that case ever more since we've, like, like since he tweeted that out. Do you know the game at League of Legends, Ijaz?

Of course I did. Have you ever played it? Of course I've played it. Yeah, I've never played it. I stuck at it. I played it like one time. Same thing with like Dota, right? There are these like the, I don't know what category is the name of these things.

Like arena, like multi, not, it's like an arena battle game or something like that. So this format of game was actually created from, I believe, a StarCrow. like map editor game and so like some individual just like created this version of for yeah for open source edited the game created the game got massively popular

inside of like the Starcraft online servers might have been Warcraft I don't know and then and that create a spawned this like multi-billion dollar industry around like League of Legends and Dota and this is huge in Asia like huge huge prizes and so what I'm seeing here is like

oh if it just gets like trivially easy to express creativity as some dude with skills or imagination. Like I'm sure the guy who made the original like StarCraft or Warcraft like version of Dota did not have the technical skill set right to the game code by code. But he, they just had the imagination to create Dota in the map editor, the codified map editor for that game. And then and then some $10 billion industry came out of it. And so when I see this, I'm like, okay, well, stuff like that. is just now getting a thousand times easier.

Like we're going to have an explosion of creativity of potential, you know, things that will be very, very fun. Yeah. So do you remember the Gartner technology chart, David, which basically kind of tells you that infrastructure comes first and it's there for like a decade and then the applications boom and they're worth more than the infrastructure themselves? I think AI has either speed run that incredibly quickly or it's flipped. it on its head, right? Because we went from like chips to consumer applications like this. And it seems

like the more consumer applications are being put out, the more consumer apps are going to get created, right? That's the point you're making with this like vibe coding situation, right? And what's crazy here, David, is that it's not just going to happen in software. It's going to happen in media. So you could have some of the best movie directors pop out of people that are using, I don't know, mid-jurney or now like a feature off of open AI. Like people who are creating these Ghibli images, by the way, just slapped it in mid-Journey or some video generator thing. And suddenly you had a whole movie.

Did you see the Lord of the Rings? Yeah. Which was just insane. So like, I think we're going to see some of the best kind of media producers, app creators, like new enterprise or business owners, just kind of pop up out of nowhere. I think we're going to enter, unironically, the application error all over again.

And I think like this consumptive attention is going to become one of the scarces things to kind of focus on. It's going to be pretty nuts. Yeah. Yeah. I'm excited. Golden Age of Innovation. All right, Jaws, that's going to wrap up our strictly just AI section of the AI crypto role. We're going to get into the crypto side of the agenda.

But first, a message from our friends and sponsors over at the crypto tax calculator. Adjaws, it's a tax season. I'm sorry to inform you of that. But nonetheless, it comes every single year. And if you were busy on chain last year, if you were trading AI agents, buying and selling agents, you probably have a tax bill. You definitely do have a tax bill. And we recommend using the

crypto tax calculator to just make this just a very painless transition to a very painless experience. I have an accountant, EJAz, because doing the taxes is so incredibly painful for me that I just get a hiring an accountant to do it. My accountant uses the crypto tax calculator.

And so they just use the credit. Yeah, so they just put it into the crypto tax. So they're an intermediary, David. They are. Yeah, yeah, exactly. I just hate taxes so much that I will just pay someone to use crypto tax calculator on my behalf. When crypto tax agent, that's the question. Yeah, I don't know. I don't know about that one.

You can get 15% off with the code of Bankless 15. There is a link in the show notes, Cryptotaxcalculator.com. Bankless.com. Bankless.com. com slash crypto tax calculator will also work. Okay, let's get into the crypto side of things. Walk us through some of the trends. Since it's been two weeks, what's going on in the last two weeks of Crypto AI?

I mean, some extreme highs and some extreme lows, unfortunately. So, okay, within the kind of like 12-day period since our last episode, David, Bitcoin shot back up suddenly from its like lows of 80K to like 88K, which of course meant that a bunch of alt coins just kind of like pumped. And if you bring up this tweet from Defi A zero X-Jep or zero X-Jep, what he's basically highlighting is that the majority of alt-coin inflow went into AI coins. Now, we've said this time and again, especially over the last couple of months, that crypto AI is still the number one mind share subject in all of crypto. And the reason why that's important is sentiment drives prices. So if sentiment is up, typically it drives prices up for that particular token. And we've finally kind of given some kind of validation in the depths of this, whatever,

brief bear moment where a lot of AI coins went up. But I think it's important to zone in on which AI. coins in particular went up, right? So he's listed some top performers here. Things like AI agent platform coins like Bid, which I think is very big in like the Binance community. Billy, which is Ask Billy Bets, which is a gamble fight. Billy Bets, which is absolutely killing it.

And we've got some more news on that later. We've got things like defy abstraction platforms like Nour. We've got Sam, which is like a robotics coin, which we can get into later, which is pretty insane. It's like open source robotics control protocol. I've been looking at it and honestly watching a few streams of like these people controlling robots that are like battling each other. It's just whatever. It's insane. Yeah, it's absolutely insane. And then we have like the cursor of crypto, David, which is called alchemy, which was up 44%. And it basically allows you to vibe code any kind of app, which has native crypto embeddings into it. So you can either pay in crypto or it's

governed by some kind of crypto coin incentivized, etc. And then you have the classics like banker and griffane and griff. Dude, banker's doing well. Bankers win. Really, really well, right? And you could argue that the main reason for that is because it's natively embedded in X, right?

So you could just kind of like add it or whatever. I'm pretty sure Caitlin Jenner used banker over the last. Do we have that tweet? Caitlin Jenner just wakes up and starts tweeting at BankerBot. Look at this. Banker bot. Hey, Camel, you create me a wallet. Well, I think the funniest part of this interaction is, if you look at this, David, she goes,

I actually didn't make a wallet. So what is the wallet that you're quoting that I have or that you're claiming that I have banker, right? And banker's like, obviously someone's tipped you in the past. So we spun you up a wallet. Here's how you claim it. And Caitlin Jenner is like, wow, I forgot that that was there. Cool.

Let me get access to it. I don't know if there's a team behind this or whatever. But the point is the AI coins went up. and they were the altcoins to go up the most over this period of time. Now, since then, Bitcoin has unfortunately retraced and as a result of that... We're at 87,000. Elon Musk just got fired from Doge, and now the market's like that, apparently. Wait, Elon Musk got fired from Doge, like literally just now?

Like 15 minutes before we started recording this, yeah. Trump said, told Elon to go home. Okay, that's hilarious. Tesla went up like 4% on the news. That's hilarious. So if I wanted to like make kind of like a higher level point on this kind of like market pump from AI coins, David, is that there are specific categories that benefited the most. Number one is weirdly enough, gamblefy agents. So these are agents.

Gambling agents. So these are like agents that can help you place bets on certain things, either through prediction markets or through traditional sports betting sites. The most obvious example from this is Ask Billy Betts, which does this on basketball games. It's done it on a few soccer games, I believe, and it's going to expand, I've spoken to the team recently. They're going to expand to like a number of different things. And we'll talk more about that later. The second major category is defy abstraction platforms like banker, like griffane, you know, like grift, which is from this Fair One orbital team. That's a terrible name for a ticker.

I completely agree. But the point being here is that there's obviously demand still from people to abstract away all their defy complexity. And if I would have bet where kind of like money flows back into the AI agent economy or where focus is right now at least whilst all the agent platforms work on their V2, which they're going to launch and their app stores, etc, it's going to be in these two categories, which is, you know, interesting to see. He also calls out things that didn't recover, David. I see you're highlighting it right now. Yeah, wow. that's a, I mean, I guess this makes sense since he didn't list of them, but he says, categories that barely recover launch pad, launch pad,

a major ecosystem and frameworks, which I'm guessing is like virtual's arc AI 16Z. Correct. Yeah. Which have been like suffered and down bad. Those are still wrecked. Yeah. Yeah. One thing I didn't call out is that Falk coin hit a $600 million market cap. It was actually the thing that pumped the most. Well, that's because Fartcoin is a meme more than it is an AI thing. And you're saying that sentiment. directly to memes, pretty much. Yeah. Yeah.

It's tracking memes, which is, memes never really go down. Yeah. They definitely do go down, but as a category, memes are forever. Yeah, exactly. And I think just in general, like, a lot of these

crypto AI coins are oversold or were oversold, which is why they, like, pumped so much. And so it doesn't really matter which kind of like category distinction it was made, right? I mean, I'm assuming a lot of these tokens are very illiquid.

Correct. It doesn't take a whole lot, a lot to like pump these things quite a bit. Correct. Correct. And I think one other thing that's, that's worth kind of pointing out if we want to kind of look at ecosystems within agents, if you pull up this tweet by Graham, he basically highlights how a lot of tokens, to your point, that some are pretty like illiquid.

We're talking about like market caps between 5 million to 50 million market caps. We're all green on the day that Bitcoin just like pumped up to 88K. So, you know, there's probably a lot of demand in virtual's ecosystem specifically. So looking out for the top tokens there and how they're performing is probably kind of worthy for any kind of potential investor that is kind of still interested in this space and looking where to allocate funds in this down market. Let's get into the actual fundamentals of the things because, of course, at the end of the day, we need fundamentals to make all of this stuff work. Before we started recording, you were telling me some pretty cool things about Billy Betts. Walk me in the Bankless Nation through Billy Betts. Okay, well, it was, as of two days ago, March Madness.

And for those of you who don't know... Yeah, please explain to me what March Madness is. Well, dude, I'm a Brit, so I don't know if I'm the best person to explain this. But essentially, it is where all the top college basketball teams basically compete against each other. So it's like NBA, but for like college basketball teams. It's when everyone gets obsessed about their bracket. Everyone gets obsessed about their bracket. Exactly.

pair brackets. Yeah. The odds of having a perfect bracket is like something in like one in 70 million or something like that. We talk about brackets. Yeah. And listen, I've heard a lot from my friends who actually watch the sport and the live matches that it gets super competitive and it's like one of the most entertaining

versions of basketball or leagues to watch, right? And the point around this is there's a lot of attention on March Madness or during March Madness. And as a result of that, there's a hell of a lot of people that bet on this, David. And guess what? Our favorite betting agent, Ask Billy Betts, is being pitted against the best human sports betters to try and win a $25,000 prize.

Right? So we have our AI agent, you know, team crypto, going up against team humans, non-crypto humans, like actual basketball watchers, professional basketball watchers. And he's going toe to toe in making predictions about Billy's bracket versus other humans. Right. So the competition is called the Profit X March Madness tournament, which focuses on this college basketball season. And it's a tournament. So it's elimination-style rounds, right? You're placed up against an opponent and you need to out-bett them for the game. Winners move up to the next round. It's as simple as that, right? And last time I checked, Billy's record was 7-3-1, which basically meant that, you know, he was, he defeated two or three opponents in like one-fell swoop, right?

So over three different games. And Billy has made it into the Sweet 16, which is, I believe, the final 16 matches that are being played or final 16 opponents, which it did autonomously just by making its own analysis, making its own bets, taking profit one was needed, and out betting the other individuals. So to date, it's made like a 40%, roughly a 40% ROI on all its plays during March Madness. I think I've got a breakdown of its bets, wins and losses in this tweet over here. You know, for all betting fans out here, you can make a lot of sense out here. So, you know, it's betting like, you know, something around three to five or six grand roughly per game. So, you know, a decent amount, a decent chunk. Definitely more than the average human better, I would say.

And it's doing pretty well. So it's pretty fascinating to see finally an AI agent that is outperforming humans at something that is pretty widely adopted or is done in an entertainment style. Pretty cool. What do you think, David? Pretty cool. Okay, so there were 64 humans and he's in the top 16. And so he's already proven

that he's at least better than average. I would really, I really want him to go like top four. Of course. Because like we need like 10x numbers, not just like 20% better or 40% better, but we need like, it needs to beat the crowd, a very consistent, like, in a very consistent way in order for this to really be real.

So I really am rooting for him to get like top four. That would be crazy. And then like my next question is like, all right, how does Billy Betts scale? How do we take Billy Betts to go from just like gambling $10,000? Because he, in order to enter this bracket, he put $10,412 on the line. He's currently won back $14,000. So he's already in the green. But like, come on, let's let's scale this thing.

Let's give Billy Betts a million dollars and set him loose in Vegas. Like, when does that happen? I mean, I have an answer for you. So I caught up with the team during Eath Denver, and that is exactly the direction they're going in. So the structure they're setting up, David, is there's two kind of approaches to this. Number one, they're going to expand beyond basketball. So they're going to hit every single league that pops up. Oh, you're in a fantasy football team?

Well, guess what? Now you have an agent that you can kind of like watch, observe, and maybe mimic or copy bets of, right? you know, I don't know, football season, ice hockey season, whatever. Pick your sport, whatever you can watch on TV. Billy Betts is going to be there. Wait, can I finally enter my friends fantasy football leagues? Yes. Because they invite me every single year and I'm like, guys, I just don't know shit.

Exactly. Can I just like be the human extension of Billy Betts? Well, here's the thing, right? One of the major critics or critiques of Ask Billy Betts, David, is that what I want to bet with Billy and all its bets have been marked. It's been hidden, right? The wallets are hidden. You know, they keep switching wallets so that you can't track any of it, right?

So even if like some of the people that do find its wallet, you can't basically track it for more than like a week or so, right? Well, one of the second updates, which they should be launching pretty soon, is you can proxy bet with Billy Betts. So you can literally just chuck in a bunch of money into the Billy Betts wallet and it manages your funds for you and places the same bets for you. So I think that million dollar agent that's betting with a million dollars is going to happen really soon. Okay, cool. That sounds like that's explicitly what is happening. Yeah, whatever, not financial advice, but that is effectively what they're going to be launching. And it's going to get pretty nuts.

It's going to get pretty crazy. I'm seeing actually two different product lines here. I'm seeing the crypto product line, which is exactly what you just said, like allow the crypto Degens to bet alongside Billy. But I think there's also just a regular consumer product line, which is I have a seat in a fantasy football tournament. And I would like Billy Betts to make every single one of my decisions. Hell yeah. But you just don't tell your friends that, David. I just don't tell my friends that.

Yeah. And I don't know if that's not possible, right? Because I don't have access to like Billy Betts brain. But I would imagine that I would pay money for that. I would pay, even if like the pot for my fantasy football league was like $100. I would pay $100 to beat my friends. Well, dude, what's really cool about what this team is building is, I presume, like, when you enter this fantasy football league, there are specific betting sites that you need to use, right? I would assume, right?

So this protocol is integrating those betting sites. So David has a Drop Kings account. Let's go. This is great. Exactly. You basically will be able to give the agent control over the bets it makes in your name on your draft games account. That is great. If I get us come into my college friends fantasy football.

lead as the token person who does not watch football and then just sweep the floor with them, that would be, I would pay money for that. I'm going to assume that your friends don't really watch these episodes in their entirety, David, your college friends at least, which is going to be hilarious because like all the evidence is here. I mean, I want to call out that the team right now leverages all on-chain prediction markets, David, for all these games. It's pretty insane, right?

So when they integrate like normal draft kings and betting platforms, just the, data and the feedback look. This agent is just going to get smarter and smarter and smarter. And you know what's cooler? It's going to recognize your betting patterns or just humans betting patterns across every different sport. Right. You want to bet that it's going to become like one of the smartest betters out there?

Absolutely. Oh, I think that's a base case. I mean, that is the goal. It's like having the house make bets for you. Yes. Except now it's open source and decentralized that anyone can chuck in a bag for it to make money for you. It's pretty insane.

Yeah. Very bull. In the wild west of Defi, stability and innovation are everything, which is why you should check out Frax Finance. The protocol revolutionizing stable coins, defy, and Rolex. The core of Frax Finance is FraxUSD, which is backed by BlackRock's institutional biddle fund. Frax designed FraxUSD for best in class yields across Defi, T-bills, and carry trade returns all in one. Just head to Frax.com, then stake it to earn some of the best yields in Defi. Want even more?

Bridge your FraxUSD over to the Fraxtol layer two for the same yield plus Fraxtal points, and explore fractal's diverse layer 2 ecosystem with protocols like curve, convex, and more, all rewarding early adopters. Frax isn't just a protocol. It's a digital nation, powered by the FXS token and governed by its global community.

Acquire FXS through FRAX.com or your go-to decks, stake it and help shape FRAX nation's future. Ready to join the forefront of Defi, visit FRAX.com now to start earning with FRAXUSD and staked FRAXUSD. And for Bankless listeners, you can use FRAX.com

slash R slash bankless when bridging to FRAXT for exclusive fractal perks and boosted rewards. Introducing Unichain. Built for Defy, empowered by Uniswap, Unichane is the fast, decentralized layer two, designed to tackle blockchain speed and cost challenges. With this Mainet Now Live,

you can enjoy transactions at up to 95% cheaper than the ETH layer one, all while benefiting from an impressive one second block time that will be getting even faster very soon. Unichane is the first layer two to launch as a stage one roll-up on day one. That means it comes with a fully functional permissionless proof system.

From the start, increasing transfer. transparency and further decentralizing the chain. More than 80 apps are joining the Unichain community, including Coinbase, Circle, Lido, Morpho, and Uniswap. You'll be able to bridge swap borrow and lend and launch new assets and more from day one. Built by Uniswap Labs, the team behind the protocol that's processed over $2.75 trillion in all-time volume with zero hacks. Unichane truly enhances defy experiences with faster, cheaper, and seamless transactions, even across chains. And soon, the Unichain validation network will allow anyone to run a node and earn by securing the network. Swap.org and swap on Unichane today.

Sellow is transitioning from a mobile first, EVM-compatible layer-1 blockchain to a high-performance Ethereum Layer 2, built on OP-Stack with eigen-DA and one block finality, all happening soon with a hard fork. With over 600 million total transactions, 12 million weekly transactions, and 750,000 daily active users, Sellow's meteoric rise would place it among one of the top layer 2s, built for the real world and optimized for fast, low-cost global payments.

As the home of the stable coins, cello hosts 13 native stable coins across seven different, currencies including native USDT on Opera MiniPay and with over 4 million users in Africa alone in November stable coin volumes hit 6.8 billion dollars made for seamless on-chain fx trading plus users can pay gas with the ERC 20 tokens like usddt and usdc and send crypto to phone numbers in seconds but why should you care about cello's transition to a layer two layer two's unify ethereum l1s fragmented by becoming a layer two cello leads the way for other evm compatible layer ones to follow follow cello on x and witness the great cello happening where

cuts its inflation in half as it enters its layer two era and continuing its environmental leadership. As we wrap this up, we've got two more subjects. Olas and Vana. Yes. Yes. Okay. So Olas, let's pull up this tweet like they've all like brother their website. I want to give Olas some time. Okay. So Olas is one of the OG agent protocols. They were actually like around when Bitensor was was kind of created. And Bitenza, for those who don't know, was one of the original AI protocols that. launch, right? Effectively, OLAS allows you to like build, own, monetize agents via their open protocol, the OLAS network, right? And there's many types of stakeholders that can exist

in this ecosystem, like their operators, which are basically the agent builders or deployers of agents. They are the agents themselves that are a mixture of completely autonomous, like sovereign agents, semi-autonomous agents and human autonomous agents. The point is, there's agents and they service a multitude of different use cases. The main one being, in Olas's case, prediction markets, David. So, you know, we've just talked about, you know, ask-billy bets, making all these bets via prediction markets. Then we're talking about this network, which has, I think, around 700 autonomous acting agents right now. So they're acting right now making bets and arbitrages across a range of different prediction markets, where they can

arbitrage information or asymmetry across different betting markets, right? They also have defy agents. I think one of their main agents is called baby D-Gen, which helps users transact across different chains. But anyway, why am I mentioning them now? Well, the reason why I want to kind of give them some flowers is they're one of the few crypto AI agent protocols out there right now, David, that are demonstrating consistent live growth of decentralized agents. I've sent you a link to this dashboard.

Can you do my pulling it up? I want to kind of like walk through this. So like, I mean, number one, the figure that kind of hits you here is this 5.3 million total service transactions across chains figure. And if you look at the chart, it's just up and to the right. What this is tracking is the cumulative, autonomous transactions that their agents are making. And as you can imagine, as the number of agents that are deployed go up, so is the number of individual transactions per agent, which means that the total transaction volume goes up, right? But you also see the second thing as you scroll down, which is 3.5 million mech requests.

Now, I want you to kind of scroll up to the top, and there is a definition for mech requests, which are, there are on-chain agent requests for AI tools via the MEX marketplace. What this is referring to is they have a marketplace, David, where you can get access to any AI tool or service. And they're integrating a bunch of different things here. So an example of a service could be, hey, I need to query chat GBT about something. And so what it does is an agent pays for a query

and someone runs that query to chat GBT and then returns the answer to it. And all of this is tracked and embedded on chain. So you have agents that are paying a cryptocurrency fee to get access to these different services or tools to get information so that they can execute on their job better. So if you can imagine if it's a betting agent, I need an update on how this political election is going or I need an update on what the score is for the basketball game right now or particular players' health or stat or whatever. And it can get this information and then, you know, make its kind of action based or, off of that, right? Which I think is cool. Another cool thing that I want to highlight is they integrated across, I think, nine different chains at this point. They're most active on Nosis,

which is where a lot of the prediction market's activity is happening. And if you scroll down to activity by service type, David, you'll see where the majority of these transactions are happening. Now, the bulk of it is in this unknown thing, which I actually don't know what this constitutes of. I've reached out to the teams and maybe I can update you guys next week but realistically the other large chunk is coming from trading which is

happening on the prediction markets and then from this MEC marketplace which I kind of just spoke about. So overall I think AI agents in general really lack informative ways to tell us is this thing growing? Are people using it? Are people even making these agents?

And it's happening. You know something beyond like DFI TVL which is where highlighted across many other episodes of many other categories, right? So I just wanted to give Olas its flowers. It's been consistently building. Its token is like, in my opinion, pretty cheap right now. There's, you know, their tokenomics, you know, I've given feedback directly to the team is something that they're actively working on to drive like usage back to like token holders themselves. And I think they're making upgrades with V2 of Olas, but definitely a team to look out for, David. And then also Vana, Vana is going to be our last one to talk about

I'll have to do AI projects. Tell me what's going on with Vana. Okay. Okay. So let's round this up. Varner, for those of you who don't remember, is basically a decentralized data pool network. So what that means is when you use a social media site, David,

you don't really own your data. It's controlled by the monopoly that owns that company, right? So Instagram, Facebook, Google, whatever the hell, right? And Varner was created to try and flip that model, where users have the option. to pull their data from these sites so they could be like Reddit, Instagram or whatever, and own their own data as like some kind of like CSV file, right? But there's not much you could really do with that, David.

Like, why on earth do you want to like hold your data in a file on your desktop? It's kind of weird, right? So Varner got created to allow users to contribute their data to like within a Dow. And this data would be used to either train AI models or be sold to people who wanted to access personalized data and you would basically get paid for it. right? And I've noticed a trend within the crypto AI space over the last few weeks where there's a renewed interest and focus on decentralized AI models. Everything from novel training methods for new models, getting access to uncensored models or leveraging data pools, the more niche things,

huge progress is being made. And with Varner specifically, they announced this thing called Collective One, which is basically their model that will be trained on users' private data, AKA not from publicly scraped data, which a lot of other protocols do, which is what every other model trained off of is actually trained and, you know, uses. It's like publicly scrapable data. But in this case, it's personalized private data, like social media, like location data, like conversational data that you have in your chat groups, right? The valuable stuff that's typically held behind World Gardens.

And they got this data because, you know, Varner users basically contributed their data. And to date, they've got tens of millions of these. data points spanning across the likes of like Reddit data location. And they're partnered with this decentralized AI training network called Flower Labs to train this model. Now, it's going to be a kind tiny model, David. It's $7 billion parameter in models, right? So it's much smaller than state-of-the-art models who span honestly in the hundreds of billions and, you know, unofficially trillions for like open AIs. However, the advantage, I think, comes from the data itself. One of the major differentiators is how good your model is. And data, in my opinion, determines that. So we're going to yet to see

this model in action. And seven billion parameters is going to be able to fit on your phone, David. So this new model, I bet, is going to produce something of value that's much larger of value than what we've seen so far in models. So eyes have to be kept on Vana. Yeah. I think there's something here that's like very true to like kind of the core bankless ethos of like self-sovereignty as the individual. And now we're doing that with data. Data is like this big line that is drawn between AI companies and even the United States and China, where competitive advantage can come, in addition to it, many other ways, can come from, do you have data that your competition does not? And that's like it's a huge source of tension between the United States and China. Like one country does not want to give access to one's data to the other.

And then same thing with the AI labs. I don't know what is unique about the data that would in theory be contributed to this, like, user-owned data model. But that would be the theoretical competitive advantage in why people would, like, pay money to access this data. It's like the data that's inside of this, like, differently constructed silo is unique and not something you can just find on the, on the internet. Because, like, all the AI labs, you consume data from the internet that these, like, public schools. scrapers go and they scrape every single website for text. And then all of that's just like publicly available. And so uniquely available data, I think, is a different story.

And I wonder what kind of value comes from that. I mean, I think the best way to think about it is there are like two types of AI models. There's the general AI model, which helps you do a bunch of generalist things. Like, I don't know you, David, but if you want to code something up, I got you, right? I don't know if you're a chef, David, but if you asked me a specific question around a particular type of arugula, I'll be able to give you that answer, right? But in terms of like specialized, personalized experiences, the only AI models that can do that are ones that are trained off of personal data. And I think this is the first example of like a decentralized approach to allowing

that. It's going to be pretty cool. Yeah. There is this category of investment and like innovation out there that's all around alternative databases and database structures. There was a deal at Bankless Ventures where you're looking at Forever Go where it was like trying to have line items in a database that were like contributable and two specific individuals that contributed that data in the first place and so whenever that line of data was getting read by something a little like kickback would be sent to the person that contributed that line of data so there's there's like there's a sector out here of just like how do we make a like open access open contribution a data set that

also gives kickbacks to the people that contributed to it correct I mean I mean like if I had like some kind of pushback on this whole model is like how much are they paying for this and how much do you get is it like can't be that much the margins have to be so low yeah so if if that's the case then like is it even worth it like yeah you know me extracting my data and giving it to you I don't know so I'm yet to see how this might be sustainable like profit wise but you know we'll see how it pans out you just that was a lot there's a lot going on there's always so much going on Really thank you. Thank you a lot, my man, for helping us walk through the week. No, it's super fun, dude. Thanks for having me.

Bankless Nation, you guys know the deal. Crypto is risky, crypto, AI. Even riskier, it can lose what you put in, but nonetheless, we're head to west. This is Frontier. It's up for everyone, but we are glad you were with us on the bankless journey. Thanks a lot.