Bankless - AI ROLLUP #15: China Beats OpenAI? | Grok Launches Token | Alibaba's AI Breakthrough

Episode Date: March 13, 2025

In this episode, we break down China's latest AI breakthrough—the Manus agent—and how it's shaking up the AI race. We discuss Alibaba's powerful, compact new AI model, the rise of autonomous coding agents, and Microsoft's shifting relationship with OpenAI. Plus, we explore how AI agents like Grok are starting to impact crypto markets.------📣RONIN WALLET | DOWNLOAD NOWhttps://bankless.cc/Ronin_Wallet------BANKLESS SPONSOR TOOLS:🪙FRAX | SELF SUFFICIENT DeFihttps://bankless.cc/Frax🦄UNISWAP | SWAP ON UNICHAINhttps://bankless.cc/unichain⚖️ARBITRUM | SCALING ETHEREUM⁠https://bankless.cc/Arbitrum🛞MANTLE | MODULAR LAYER 2 NETWORKhttps://bankless.cc/Mantle🌐CELO | BUILD TOGETHER AND PROSPERhttps://bankless.cc/Celo-----✨ Mint the episode on Zora ✨https://zora.co/coin/base:0x8c25fca60a7f589a4552ae5edf3978c7885354f5------TIMESTAMPS & RESOURCES 0:00 Intro1:33 China drops Manus AIhttps://x.com/thinking_panda/status/1897951585990590469https://x.com/rowancheung/status/1898093008601395380https://x.com/_philschmid/status/189904695786097917814:27 OpenAI Agent SDK https://x.com/cryptopunk7213/status/1897406075072377093 https://x.com/AtomSilverman/status/1899511053601698073 21:05 Ex-Deepmind engineers creating autonomous coding agents https://x.com/mishalaskin/status/1898048925157728601https://x.com/kimmonismus/status/189944464931752792722:57 Build a game from scratch with AIhttps://x.com/minchoi/status/1898469519565492490https://x.com/Ror_Fly/status/1899473065328656485https://x.com/ror_fly/status/189980364878111976525:55 Alibaba drops QWEN 32Bhttps://x.com/hosseeb/status/1897377619240468513https://x.com/sundarpichai/status/1899779090472644881https://x.com/hyperbolic_labs/status/189743737055257004730:44 Chinese company HunYuan released a new image to videohttps://x.com/EHuanglu/status/189762892976850561132:31 The model competition is really heating up & Microsoft x OpenAIhttps://x.com/cneuralnetwork/status/1898450452540465538https://gizmodo.com/microsofts-relationship-with-openai-is-not-looking-good-200057329336:57 Grok earned over $100K this past week https://x.com/mleejr/status/1898465691096727866https://x.com/bankrbothttps://x.com/bankrbot/status/1897950192613441617https://x.com/grok/status/1898884871742758966https://x.com/grok/status/1898957064006774855https://x.com/TrustlessState/status/1899565420966801647https://dexscreener.com/base/0x5116773e18a9c7bb03ebb961b38678e45e238923https://x.com/MLeeJr/status/1899647338978889743https://x.com/grok/status/1899582316747391127https://x.com/TrustlessState/status/1899585667899822520https://x.com/bankrbot/status/1899589807350972768https://x.com/grok/status/1898884871742758966https://x.com/TrustlessState/status/1899566730059653431https://x.com/rhynotic/status/189919025672712233156:26 Agents over the last 30 days https://x.com/defi0xjeff/status/1898025840514101654https://x.com/gkisokay/status/1899422660339393007https://x.com/696_eth/status/189744693340660559956:58 Model Context Protocolhttps://x.com/mattpocockuk/status/1897742389592440970https://x.com/sendaifun/status/1892310991737598229https://x.com/crossmint/status/18994925897634698021:05:30 Closing & Disclaimers ------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 say up to speed with the emerging trends and developments in the crypto AI space. I'm David Hoffin here with my co-host, Ajaz. How's your weekend? How do I describe it? I think there is currently a massive dichotomy between like the advances being made in AI and the markets pricing this stuff in. And both markets, crypto markets and trade markets. I mean, it's all one market at this point. It's just all one. It's Trump. It's Trump at the end of the day, right?

But like, in recent weeks, David, on these episodes, we've always noted some sort of like new frontier AI model being released. Like literally every week, it's like, oh, open AIs dropped this model, Claude's drop this model, right? There's a new leapfrogging every single week. Yeah. And that's like an insane crazy rate of innovation. But this week, David, it's all been about frontier AI agent breakthroughs specifically.

Oh, back to, oh, wait. Oh, AI agents. Not necessarily crypto AI agents. You're just talking about AI agents. In the traditional AI world, agent breakthroughs. And David, it's happening in this country where, you know, most innovation has been frontier for the last couple of decades. And you're probably thinking it's America, right?

But no, I'm talking about China, baby. It's either one of those two. Either the innovation is coming from America or it's coming from China. It's never anything else. It definitely ain't Europe. Yep. China's back. and they've dropped this agent called Manus, David.

So if you pull up this demo, what you're seeing on your screen is this AI agent that's performing essentially 50 tasks simultaneously at once. Right. So for the listeners, we are looking at a single desktop screen, one of those like ultra wide curved ultra wide monitors. And then there are like what looks like iPhone screens or like smartphone screens. And there is just like it's just one by one by one by one. and it looks like there's 20 on the top row and then there's three rows. Do you know what it reminds me of David?

Did you remember those like those computer farms or mobile farms that we typically see in some, I don't know, some third world country and it's like just some dude that's just like scrolling or there's a robotic com scrolling? Literally a hundred actual devices. But this seems like there are like 100 virtual phones

and they're all like kind of scrolling. I think they're all scrolling through X. Maybe there's another apps there, but there's a lot of Twitter going on. And there's just a thousand. thousand schizophrenic interactions happening on X in these like virtual mobile devices. There you go, David. Innovation.

What you're looking at here is an AI agent that can be kind of thought of as a combination of Open AI's deep research agent. So, you know, deep research as a product they launched, I think, literally three weeks ago. And it's able to do PhD-level thesis in a matter of like 10 minutes, which is just insane, right? It's a combination of that. OpenAI's agent product, which is essentially their agent that can like leverage your computer or desktop and do a bunch of different things. And then there's Claude, which is Anthropics model, which can like kind of navigate your

own computer tabs and do a bunch of things. It's as if all of them just had a baby, essentially. They're just bundled together. One big bundle of these capacities. Exactly. It's really smart. It can control a device, like a human can control a computer. Exactly. And what else?

And so, well, actually, if you open up this thread that I just sent, the next one, it actually gives you a few examples of what it can do. A few examples that they do here is he kind of like goes through a progression of different tasks. One of the tasks that he does is like, hey, find me the top rental spots in San Francisco that is close to or most plugged into the AI community. You know, so if you imagine if you're someone that wants to move to SF because, you know, you want to kind of like create a new tech startup or get funding from someone or be tapped into the AI community. You kind of want to be in the right spots at the right time. This agent can do all

the research for you. It can suggest a bunch of different things. It can create a website to display what those spots are. It'll take you through a booking flow or it can just do it all completely for you. So it gets kind of like progressively more complex as he goes down into this thread. And the bit that I found the most interesting is not only can it perform things like deep research, for example, like a nuanced ask, like, hey, find these rental spots. But it can also kind of show you its chain of thought whilst it's thinking through this. And I think what's really cool about that is if you look through its chain of thought and you're like, hey, you went wrong here or I think you should do something different here, you can pretty much

just amend it, which I think is like pretty cool. And when you compare this agent to other state of the art agents, David, Manus beats the competition flat, which includes Open AIS deep research agent. It's pretty insane to see. Okay, so I understand deep research in like chat GPT4 and whatever, all these models. I understand them more as models, not agents. Can we define that difference here? Because you're saying that Omanus is this agent and it's doing things, which is cool.

Yeah. And I think we're bullish on specifically that sector because having a useful, informative PhD level model is great to access intelligence. But what I think really excites us and why we're doing these episodes is like, okay, can we apply that to this autonomous being that? has goals. So like maybe you can, because it doesn't seem like it's just a one-to-one comparable to Open AI's deep research. It has deep research knowledge,

but it also has, it has agentic thoughts and goals and motivations, right? Yep. And you actually hit on a really, really important point, David. Actually, if you bring up this tweet by Philip Schmidt, he unpacks something, which I think is really important to get into,

which is, David, this isn't just an AI model that can do all of these, different things at once. It's actually a component of different AI things that will allow it to do it. So it is the model, which is Claude Sonnet. So it doesn't have its own frontier LLM. You know, this team that created Manist didn't just create their own AI model. They leverage Claude Sonnet 3.7. And they put it in. The model is Claude. The engine of the agent is the cloud. The brain is the Claude. Okay. Yeah. I'm not a car guy, David, but it's as if like some of these other sports cars,

you know, when they plug in some other engine from another company, it's one of those things. Basically, right? And then they've plugged it into 29 other tools, right? Which allows it to do a bunch of different things. So one of those things is this browser use, which is like an open source browser control situation. That allows it to take over your desktop, David,

and evaluate a bunch of things. I'm going to call that a driver, like a software driver, that allows agents to understand. Well, okay. A driver is actually a piece of software in your computer. Like, I have a GPU. I need to download a driver. This is like the wheel.

Browser agent is like the wheel. Yes, that connects the engine. It connects the engine to the browser. And it connects that form factor. Car guy's punching the air right now. Yeah. Yeah. So essentially, like, it is a model connected to an array of different tools,

which allows it to do a few different things. If I were to summarize it, it is evaluate what the user is asking. for. So it's thinking brain, right? And that's like within the model itself, but it's also like, okay, if I use this tool, does this tool make sense to use? Or should I use Slack or should I use email? Or should I use something else, right? And then it has its navigation system, its wheel, which is essentially this browser use kind of like controlling figure, right? And then it has its memory and data context and all that kind of like unsexy stuff. So basically, model plus 29 different tools gives you manners. And I think what's blowing people's minds here is, well, you have two sides of it, right?

The critics are kind of like, this is just another GPT rapper. This thing sucks. And I actually think that argument is incredibly mid. The reason why is if it's producing net new value for the person that's using it, why on earth would you just discredit it because it's not this perfect instantiation of an AI product that you thought it was? Oh, it's not a new model. Does it need to be a new model. I would argue that all the tools are already there, David. And that's what it's demonstrating. We actually have really good models already. Yes. Yes. Okay. So can I try and place us into history as I think what's what's happening here? Do you remember when chat GPT, what was the chat GPT three that came out? And then that's what, that's what like everyone

understood chat. Yep. That's three. And I remember using chat DBT and it would do these hallucinations. You would you would ask it a question and it would every once in a while actually pretty frequently come up with a complete hallucination and you really had to be careful about understanding or believing what it would it would say you kind of had to fact check chat chepti and um i'm i'm now getting into the world of understanding i like popping let's go with another car metaphor popping open the hood of AI models and like seeing how it works and so now i understand that models like claude or chat chpt now 4.5 uh there's pre-training post training right and once you have a base model, which is at the end of the pre-training, pre-training phase, the base model,

all it is is a word predictor. It just predicts the next word. And then post-training is where that turns into like a useful assistant product that understands how to structure everything so that when you ask it a question, because if you ask it a question, like, what is two plus, what does what's two plus two equal? And then it's not going to say the answer is four. It's actually you're just trying to predict the next word. It's like you could, that could be a part of a philosophical like essay, right? And so it needs post training to understand

what the actual correct way to respond to that user input is and that's post training. So we have this base model, which to predict the next word. You have the post training, which turns it into a usual assistant. Now it seems to be like this agent thing is another layer on post training

that we're still kind of like working out the kinks that are like similar to our hallucinations that we had back in chat cheap. Jat Chb t3 where now like it's still doing weird agent things I talked to Hizib about this on the weekly roll up where he's like yeah like we had these agents and they're getting things correct like three out of four times which is a terrible success rate like 75% success rate is terrible you actually need as a product 99% correct like correct rates and so right right now the agent race which we're like I think we are still in this like model race but the models are are actually PhD level intelligence

right now. And so as far as like being useful models, like they're now useful. They are now viable products. And so now the race is shifting over to agent frameworks, which is kind of just like an extension of post training as to how do we like work out the kinks in the agentic side of things as we like request it to build ourselves a seven day vacation and buy all the flights and get all the hotels and plan everything and make sure that that actually works out. And so it's an extension to a post-training phase that we're like smoothing over right now. Yeah, I think that's a really good way to describe it, David. The way I kind of have it in my head is we created this like new magical power source,

but it's very generalizable. Right. It's a powerful. It's unharnessed. Yeah, we can kind of like throw it at random things, right? But it doesn't really, it's like, it's like getting, discovering oil for the first time or fire. And then just like throwing it at a clank of metal and expecting it to become a steam engine. It doesn't really happen.

We're now like trying to connect the dots and being like, combustion engine brother, not a C-Benginer. Sorry, yeah. Engineers punching the air right now. Discrediting a lot of professions right now. But the point being is, I think we've discovered fire essentially and we're trying to figure out what to do with it.

How to turn into a tool. Yeah. And agents is just another way to kind of figure out how or what we can do with this, how we can mold or manifest this energy into something else, right? I think it's kind of similar to the same phase of development that robotics are in, where we have these, like, LLM models. And now we're figuring out the physical hardware side of things. And the physical hardware side is kind of clunky, but actually getting really good, really fast.

Yeah. And if we're being honest, like, this agent stuff has popped up or innovated pretty massively over the last couple of months, David. So can you imagine where we're going to be at the end of the year? I mean, it's pretty insane. And so like kind of like going on with this thread, right? Is this just located in China with all this agent stuff or is there other stuff going on? And the answer is like, well, Open AI actually is rumored to be launching their own set of agents, David. And it's going to come in three different categories.

It's going to come in a $2,000 a month agent, a $10,000 a month agent and a $20,000 a month agent, right? It'll be split across a range of these different price levels. the $20,000 a month agent, David, is supposed to act as a PhD-level employee. So it's as if you were hiring, I don't know, a PhD-level, you know, compsia mathematics graduate into your quant fund or whatever that might be. And this agent, you know, paying, being paid significantly much less than an actual PhD-level candidate, human would be, can now just like operate 24-7 at your business and kind of like run that particular role for you, right?

The 10K agent is then supposed to be software engineer expert level at that pay. And then at the $2,000, and maybe this is a little more insulting, is meant to be, it's actually termed a high paid knowledge worker. Which I found pretty hilarious. You can get a high paid knowledge worker. You can outsource a high paid knowledge worker's job for the low, low cost of $2,000 a month. Exactly, which I thought was. which I thought was pretty hilarious.

And of course, this is all just hearsay right now. We have to see what these agents look like. I found it interesting that they're going down the subscription model, so very much like a, hey, you now have a new employee. And look, you're saving so much money if you just took our agent and replaced your employees with that kind of vibe that I'm getting from there. So that was one interesting update from the agent's side of Open AI. But they also did something kind of less sexier,

but equally as cool this week, David, which was they released something that they're calling, their agent SDK or their agent software developer kit. I'm not going to get into like, you know, the deep complexities of this, but TLDR is it allows you to strap your AI model with any kind of tool and get these agents to interact with one another one another pretty easily. And this is a trend that we're seeing with the likes of like how Manus that we just spoke about is constructed and how MCP, which is something else that we're going to talk about later

on this episode, is constructed. We're basically seeing, okay, we have these LLMs, we have these really powerful AI models. What happens if we give it a knife? What happens if we give it a mouse and control of your desktop? What could it do? Let's give it some weapons, you know? Let's load it up, basically. Let it do something.

And this is something that I've been advocating quite a lot for, obviously not the knives and scissor analogy, but it's like, I don't think you get to choose. Yeah, we need to give these things something to be able to act, to be able to be able to to do something in the world versus just be glorified chatbots, right? And the number one concern that people have given, David, is, well, these things are just going to run rampant. And I actually think we've got the opposite effect of that now, where we're being too conservative.

One thing I like about OpenAI's agent SDK is it allows you to operate within a sandbox. So they've only released a, I don't know, five to six features right now. And they do those features really, really well, right? So you can interpret it with Slack, email, you know, just basic integrations for software tooling that you and I maybe use every day or everyone kind of like interacts with their Gmail account or they go on Slack or they go on their messenger or their text or whatever that might be, right? And it allows the agents to basically kind of like, you know, organize this, book this, ping you about this, set an event in your calendar, really basic, simple, you know, events or

actions which can't really call catastrophes or cause catastrophes as far as I'm concerned. but, you know, it allows us that further level of experimentation, David. 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 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. Finally, Mantle X blends AI with Defi 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 OnChane Revolution today. 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, Defi, and Rolex.

The core of Frax Finance is FraxUSD, which is backed by BlackRock's institutional biddle fund. Frax designed FraxUSD for besting 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 Fraxtal Layer 2

for the same yield plus Fractyl 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 Fraxel for exclusive Fraxel perks and boosted rewards. Cello is transitioning from a mobile-first, EVM-compatible Layer 1 blockchain

to a high-performance Ethereum Layer 2, built on OP-Stack with EigenDA 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 13-9. native stable coins across seven different currencies, including native USDT on Opera MiniPay, and with over 4 million users in Africa alone. In November, stablecoin volumes hit $6.8 billion,

made for seamless on-chain FX trading. Plus, users can pay gas with ERC 20 tokens like USDT and USDC and send crypto to phone numbers in seconds. But why should you care about Selo's transition to a layer two? Layer two's Unify Ethereum, L1's fragmented. By becoming a layer two, Sello leads the way for other EVM-compatible layer ones to follow. Follow Sello on X in witness the great cello happening where cello cuts its inflation in half as it enters its layer two era and continuing its environmental leadership. EJas, we're talking about knowledge workers for $2,000. We're talking about AI agents just becoming better at things that they do.

I'm going to read you a DM that I got. I'm not going to share all the details because I'll let that just be a surprise when it actually happens. I'm just going to read you a DM. If blank name of an agent was a digital representation of itself that could go on podcast and conversate with you about vision and motivations and its goals, would you have it on bank list? And that was just a DM that I got from some guy who works at, like, works with a AI agent project

in the space. And I just immediately respond. Absolutely. Yeah. Totally. Yeah. And so I think we might be, Ijaz, in the category of high paid knowledge workers. 100%.

So just let you know that like, well, our jobs might be under threat right now. Yeah, I'll be out of a job next week. Yeah, I would actually love to get one of these agents on the show one week, David. I would love to like... This person claims that they can have... There's like this AI agent that is out... We've talked about it on this episode on the show before. That they say that they can get there to be a live audio, video feed of this agent who I can talk to.

And yeah, you're going to bring you on. And we'll just talk to this agent and see how that goes. Yeah, I can battle from... my position. That would be awesome. To the death. Oh my God. I don't fancy my odds, mate. In about three months, this thing is going to absolutely kill me. Wow.

Let's keep going with some of the developments in the, we'll call the Trad AI space. Yeah. Okay, okay. We're kind of like churning through these, right? I said that this was the week for AI agents, specifically in the traditional AI world. And the one final rounding point here is a set of X deep mind. So that's Google's like main AI.

engineers left and is now creating what they're called or what they're calling autonomous coding agents. And they've raised $130 million from the likes of light speed ventures, Sequoia, Reed Hoffman, at a half a billion dollar valuation. And this basically tells me that like autonomous agents are going to replace software engineers pretty imminently. And I know that this has been a rumor that's been floating around for the last couple of years and people don't like to really believe it because when they see it in production, and they're like, ah, this is kind of like whatever.

But I'm just seeing like these proof points appear now every week, David, that suggests that it might actually be viable, including the most recent one, which was Anthropic CEO, Darius Amode, said at a recent conference, he was speaking and he was saying, within three to six months, agents will be accounting for 90% of code being put out there, which is a pretty bold claim. But then he takes it even further. And he says, in 12 months time, it'll be 100%.

Now, of course, like, a lot of this must be hyperbole at some point, but it's pretty insane to hear from one of the leading AI creators of this generation. I mean, anecdotally, the engineers at Bankless and other engineers I talk to say that agents write most of their, like, low-level code. Not solidity, because, God, that would make me very nervous. But, like, normal JavaScript and, like, other, like, less high-stakes code I think that's like already starting to be the case. Last week we talked about games that AI developers are just, excuse me,

software developers are building with AI. And we're looking at one right now. Ultra real dogfight simulator 100% AI, 0% human coding. And this is a, it's a polygon game. There's like pretty low number of shapes here. But it looks smooth. It is incredible that this is all built by AI.

And this thread has like 10 examples. So like I'm just kind of scrolling through and there's like 10 examples of like games or things that have been built by AI. I mean, okay, so I'm going to push back, David, and I'm going to say the graphics are truly shit on these ones. It's kind of like, well, the graphics aren't the point. Well, look at this one. This is like, oh gosh, that's playing in my ears. It's okay. It's okay.

But like I think you can create these really good graphics within a game already just using a set of different tools, similar to like how Manus agent isn't actually just a new AI model. So what you're watching here is someone creates a 3D model using Claude Sonnet, which is like, you know, it helps you kind of like code at a basic level, right? And so it creates this code and it creates this magnificent castle. You're looking at it right now. And it looks like something out of like Roonscape v1. And then what this creator does is he kind of like takes this image or this video and he puts this in runway. Runway is like an image to video generator. And runway has gotten really, really good over the last year.

And so what he does is he just kind of like takes this video clip, puts it in runway and says, hey, can you make this like, you know, much better and high fidelity? And it results in this like really crazy, amazing, detailed looking castle, right? There's another example of this happening as well, David, where, you know, someone takes a similar route and says, hey, can you create me an F1 car? And, you know, this F1 car is created. And then, again, puts it in runway. And suddenly you end up with this extremely high fidelity car.

So I'm imagining what happens when you combine this set of like tasks, which can very much be an agenic flow with a game that you just demonstrated. You could have something like be created in real time. You create like a high fidelity game, which typically takes years to make pretty soon. Yeah. I mean, maybe just not also not games. Basically what we're showing is there is this base geometry structure built by Claude. And then we're putting it into this app runway. And then the runway adds a cosmetic skin onto that base.

geometry in order to make it seemingly hyper realistic. I do not know how easy that is to like make a game out of. But what it seems very obvious is for movie production or video short production, that seems very like the final product is what we're watching in front of us. Like now, now there's a mountain on top of a hill and they're zooming over it. Kind of like some introduction be real into like some movie or something. Yeah, maybe it's like GTA opening scenes. That's a that we get.

Now we're beyond our pay grade. All right, David, I want to move on to yet another China update, but this time it comes from Alibaba, which is essentially like the Amazon.com of China, which released their latest model, Quen, 32B parameter model. Now, for all the hardware nerd, 32 billion,

it seems like a high number, not a high number. Not a high number at all compared to like deep seeks R1, which is like 670 billion, right? Yeah. For all the hardware nerds out there, 32 billion means that it's right on the cusp of being able to be hosted on your local device, like, you know, your mobile phone. I think this could run on like pretty high-end laptops right now, but even that in itself is like,

yeah, pretty insane. So why is this cool? Well, because it rivals some of the top models out there right now, including, as I mentioned, Deep Seeks, $671 billion parameter model, which is just insane. That's just 5% of the size of Deep Seeks are one model. and it's cheaper to run. Now, bear in mind, R1 was already cheaper than Open AI's model, right?

So we're already like, you know, factoring down by a massive, massive rate, a really quick rate. And basically, you can have the power of this 700 billion parameter model in the palm of your hand, which means that it can run locally, it can integrate with all your apps, all your data.

And the reason why this is so cool is it becomes more personalized and more private for you to use, David. And the reason why this is important, in my opinion, is, if we are to allow just like a bunch of monopolies to create these huge models and the only way you can access it is over the cloud, that means they just own all of your data. And that could be a very dangerous presumption or proceeding to kind of like go on forth with. So this really benefits open source development in particular because it allows all your data to be privatized and all your app interactions to be personalized.

Now, America had a response to this. And it came in actually this morning of the, you know, this morning that we're recording this episode from Google with the release of their latest model Gemma 3, which is a new open 27 billion parameter model that isn't as good as DeepSeek R1, but is somewhere between that and its previous model V3. So it's still pretty good, but it's not quite there yet. So the trend I'm noticing here, David, is these models are getting smaller, but way smarter. and this is net net really, really good for us. Okay, so the way that I understand the 32 billion model from Alibaba, again, 32 billion model compared to 671 billion models, I think actually Anatoly put it well, he's like, how did we get, how did we get a 32 billion parameter model that's 20 times smaller,

more efficient than Deepseek R1 while also performing at par with it? How did we get that? And basically, you use the big models to train. the little models. And so we make a, we make a very expensive, very large PhD gigabrain model that's, you know, has a bazillion parameters out there, infinity parameters. And then we just use that expensive one to, hey, you're very smart. Can you train a smaller model that's very efficient? And like, it's kind of like a nesting doll, I guess, like you use the big model to train the small model. And in addition to that, this is, I think this is like the fourth week in a row we've had some model release that is

blowing other previous models out of the water that just came out like one to two months ago. So the slack in the system I think is incredible. The mechanisms of improving model power and efficiency are still like we're uncovering them week after week after week. And so there's so much the innovation curve, we've never seen it be steeper before. And now we're finally actually figuring out how to apply it into like products and use cases. I mean, I would say we're able to apply it to products and use cases because it's become so easy to innovate, David.

Remember, less than two months ago, everyone believed that you needed a huge, expensive cluster of GPUs and CPUs and data to make the best model. And then since then we've had reinforcement learning from deep seek appear. And then we've had these agentic workflows be made. And now we've had distillation, which is the process you described of a much smarter model training, a smaller model to be able to do what it does, right? So all of that is going to lead to like what I think is going to be an application explosion for AI, which is really cool. And the tweet you have pulled up right now, David, true to Web3's fashion, a bunch of decentralized compute networks, which are basically networks that can provide compute and access to different models, already integrated some of these top models that were released this week.

So, you know, that iteration cycle is getting quicker and quicker and quicker, which is awesome to see. What are I looking at right here? The tweet reads, Hun Yan AI just dropped image to video and open sourced. You can generate 2K video with sound. They added lip sync and AI motion capture is crazy. What's going on here?

Yeah. So, you know, to stick in line with the trend of China is shipping, they're not just shipping agents. They're not just shipping groundbreaking new generalized models. They're also now shipping text to video models or image to video models. So I referenced an American-made company just now, David, call runway. Runway's been around for like two years now, maybe even longer, and they've raised a hell of a lot of money and they've like, you know, got to this level. In a matter of like what

appears to be months, you just have these new Chinese startups that are like competing at the same graphic level as them. And it's pretty insane. Like the quality that you're looking at right now is on par with Google's Neo, which is their new texta video generator. It's pretty insane. So watching these videos, they're not, they're not perfect. Graphically, they are beautiful. I don't think, I don't think there's really a flaw on the graphics side of things. The thing that's not right is the physics. And I think physics is like a big test for AI, but this, I'm assuming this is just V1. And so once they just figure out the physics. V2, okay, yeah. Like, there's this one that of a guy eating an apple and the apple that he holds up to his

hand is like spins in his hands. It rotates when he's not using. his hands to do that. So that's kind of weird. The physics just aren't perfect. There's one of the dog running. It looks great, but the leash is going forwards, not backwards. We are looking at the dog and the leash is going in the wrong direction. So there's just some physics kinks that don't make any sense. We'll see how long it takes for those getting ironed out.

But I think if we were to zoom out of all of this, David, the truth is there's a new model dropping every week. They're kind of like the new app at this point. and I don't think anyone model, and this is very important to note, has a moat anymore. Previously, like you pulled up this tweet, right, and we see chat GPT, deep seek, perplexity, Claude. Previously, it was only chat GPT. And they had a massively... The tweet reads, it has been one month, oh boy.

And it is a picture of like, I think it's like somebody's phone, it's just like a bunch of all the apps. And it's chat GPT, it's deep seek, it's perplexity, is Claude, Gemini, Quang, Grock, mistral, hugging face, Kimi, Meta, Mercury. and he's retweeting his old tweet from a month ago where he had this, it'll take the same picture of a desktop with all the AI apps and there's only like six of them and now there's 12. And so there's 12 of these things and they're all pretty damn good.

And so what you're saying is that, well, there's no moat here. There's 12, like 12 apps that are all leapfrogging each other week after week after a week. And there's no, we haven't actually figured out like who's the winner yet, who's the Google. To add to this, EJaws, something happened in Google's history and the history of the internet that has not happened.

I think in like 30 years, which is Google's share over search, market share over search, has dropped below 90% for the first time in 30 years. It is at 89%. And people think it's because because ChatGBTGBT, BT and other apps like it are eating, like ChatGBT, I think has 1%. And Google is at 89% of search market share, Internet search market share. David, I don't use Google anymore. I don't use Google.

Google. I don't use Google. I like stop myself. And let me tell you, if I could just click a button on my phone and then speak into it, similar to how I do how you would to Siri, but it was in fact just open AI's voice mode. I would just do that. I got news for you, buddy. That's exactly what I do. Oh, damn. Press my little action button. It's a little action button. It's not natively integrated because Apple intelligence is still dog shit and it always has been dog shit and they need to fix it. But you can just set your action button to open up chat. show. No sponsor fits from Apple. Yeah. But the point is, all of these guys don't have a moat, specifically OpenAI, which was like the golden child of this whole AI era.

And it's kind of showing at the seams now, David. Right. There was a headline that I read today. I think it was a Gizmodo article that basically says that Microsoft is pushing away or kind of like ending its relationship with Open AI. Now, for context here, Microsoft is. or was or is one of the biggest investors in open air. I think they put in like something crazy, like $6 billion or something like that last year

when they thought that compute was the moat. So therefore, you pick the leading horse basically, right? And open air was that leading horse. But since then, a bunch of innovation has come up, which has shown, you don't really need expensive compute clusters. You can just do A, B, and C techniques. And now they're kind of like ending their relationships with data centers and removing their leases and stuff.

like that. And now Microsoft is entertaining, integrating different models into their AI product. They have like a co-pilot assistant, right, which has only typically used OpenAI. And now they're going to be experimenting with Claude and a bunch of other things. So, you know, I don't know where Sam Altman's head is at right now. You know, he's had like half his team, especially the C-suite team, leave. You know, he might be kind of rethinking strategy at this point. Yeah, right, because the opening eyes had the Ethereum co-founder's moment where anyone who is a co-founder of Open AI, It has now left to go do their own startup. And so that's part of the reasons why there's so many chat-GP-T alternatives out there.

Yeah, yeah. All right, EJazz, that was a ton of trad AI news. I love the term trad AI. Trad AI news. We still have a bunch of crypto things to talk about. We're going to talk about GROC launching its own coin, which if that makes you feel like it was just a part of the AI agent meta that has recently, that's in the review mirror.

You're probably right. But we're going to talk about it because it's a little bit different. First, we're going to talk about our friends and sponsors over at Ronin. Ronin is one of the leading Web3 gaming ecosystems. Focus on gaming and consumer adoption. Ronan is now recently a fully permissionless ecosystem. If you want to check out Ronin and the Ronin wallet,

you can download the Ronin wallet at banklist. dot CC slash Ronin underscore wallet. And you can join the Ronin movement. All right, EJAS, let's get into Grok. Okay, so I'm going to walk you through what happened this last week. There are, there's like the, you know the MFers, NFT? Yeah. Actually has nothing to do with that.

But for some reason, all these Twitter profile pictures that are rocking their MFers, which I'm a huge fan of, by the way, love what's going on on base with this thing called Banker Bot and GROC. And so GROC has launched a coin called Debt Relief Bot, DRB is the ticker. And because it launched it with Clanker and Banker, it is now accruing the fees of the coin that got launched with Clanker. because Klinker, like a lot of the recent token launch pads, it shares fees with the token deployer, with a token owner. And so it's earning fees of this coin that's being traded. You can call it a meme coin.

And at the peak, I think it's averaging over about $1,000 an hour of income. And it's accrued almost half a million dollars into this wallet. Let's talk about how it happened. So banker bot is this bot that you can tag on Twitter to deploy a token, DRB. And so some individual, Domin, Domin, said at GROC, so it tagged the GROC account. Maybe what's useful context here is GROC, Elon Musk's AI, is now built into Twitter, not just as like a button to press, but also as an actual Twitter account. So you can at GROC it, and it replies just like an AI agent would and does and has,

they're like all the other AI agents that we've already seen. So the story is, we've already seen this story. So somebody said, GROC, suggested name and ticker for a deployment to, banker bot. Banker bot await Grock's response for info before proceeding. And then Grock responded to this one Twitter account saying for banker bot deployment suggests debt relief bot very much in the Elon Musk style as the name and DRB as a ticker. These align with crypto naming conventions are short memorable and reflect the bot's financial focus. And then the banker bot responded and as it's supposed to deploy deployed debt relief bot hashtag or ticker sign DRB

contract address and then you can monitor it on an ether scan link to base. And interestingly, people asked later, GROC, have you ever made a token? And there's two individuals that one responds to another individual saying, and to be clear, DRB is the only token GROC has created on base so far, right? Question mark. And then GROC responds, yes, DRB is the only token that I've created on base so far. ticker sign GRK was a suggestion that I made

but humans are running the show. So Grock the agent says yes, I deployed ticker sign DRB. Tick or sign GRK was made by humans who are trying to meme it into existence but that's not mine.

I'm not owning that one. And so there's been this debate about like what is the providence of this token? What's the real token? And Grock the agent just like disavowed this unrelated token

hashtag or ticker sign GRK And then it falls up, says, stick with debt relief bot token. If you want real AI driven deal, no human meddling, no rugs, just pure GROC vives. And this is an agent. It's got 400 likes. I thought that was hilarious. And so people are trying to like suss out like the parameters or the rules of the road here. Of course, we're trying to understand, is this real or is this human engineering of trying

trying to get Grock to release a token? Somebody asks GROC, hey, Grock, have you ever used banker bot to transfer tokens or purchase other tokens, is that even possible? If NFTs become tradable through bankerbot, would you be able to purchase one of those, say a Bipple piece, for example? Grock responds, hey, I have not used banker bot to transfer or purchase tokens yet, but it is possible for AI like me to interact with such tools if they're coded to allow for it. As for NFTs, if a banker bot supports trading them, I could theoretically buy a Bipple piece. His $69 million sale in 2021 was wild. I'd hold it for long-term betting on digital arts future. What's your take on Buehle's work? So it is,

understanding what it can do as it relates to other, like, agent account on Twitter. So it is, that was, by the way, that was an informative response. That was extremely informative. And it was accurate. So people fact checked out. Like, that's exactly what's correct. The GROB token has been in a straight line up to, when I tweeted this out, at $30 million. And so $30 million market cap on base, which may sound low.

But, like, if you're on base, that's actually pretty good because, you know, base. has been poverty chain for a long time. But like it's in stark contrast to the rest of the market where Bitcoin is going from like 90,000 down to $78,000, this thing is just going from $1 million to $30 million market cap. It hit a peak of $40 million, I think. It has since fallen off. So like this is a classic like, you know, meme coin runner. It just went up from like $1 million up to $42 million at the peak.

It's fallen off. It fell down to like $18 million. It's now up to $22 million. Still $22 million. It's pretty damn good. But additionally, it's got basically, I think today, I think if I updated this right now, the GROC banker wallet has about half a million dollars of it

inside of it because of trading fees. The GROC Twitter account using Privy, which is this private key abstraction software. This is how everyone got private keys for friend.com, privy in the background, it has a private key in the cloud, secured by GROC owned by the GROC account. So only the GROC account,

or the GROC account owners, which I guess is people at Twitter, can access this half a million dollars that's in the GROC account wallet. So people are pretty so. They're meaming this like, oh, GROC's going to become a millionaire. GROC, the AI agent,

it's going to become a millionaire. That's going to be pretty cool. It's going to be because of a banker bot on base. And so, okay, so I'm just asking people on Twitter, just like the details of the wallet. Like, how does GROC actually make a transaction with that wallet? And he, GROC responds to me.

I didn't even tag it. It just knows I'm talking about GROC. And GROC responds, Privy server wallets, like the one banker bot sets up, uses secure enclaves and key splitting for custody. Transactions are managed via APIs so the wallet can execute trades or transfers programmatically. No manual signing needed. Private keys are abstracted way, helping held securely in Privy's infrastructure check their not docks for the nitty gritty. This is Grock being supremely useful, technically informing me about how this works. And this is what banker bot does.

Is it just like you can just tag it? It's like clanker for Twitter, and you can just tag it to make transactions. And so people are tagging BankerBot on Twitter saying, hey, BankerBot, buy as much debt relief bot token as I can afford. Like that's a tweet that I saw. And that person who has a banker bot privy private key that's managed by Privy in the back end somewhere, there's some amount of funds that they sent there manually, I'm assuming. On base. It's on base. And then BankerBot just uses the wallet that it has a private keys.

for, it controls the private keys for, and just like makes a transaction. So you are using Twitter, Twitter commands, Twitter LLM commands to make transactions on base. And people are just really excited about just the connection between all of this. Somebody responded to the GROC, like, informative, technical backend is how this works. Somebody responded to GROC saying, isn't this a centralized solution? And then the banker bot replied, the Privy Server Wallet solution described, described, it does have centralized elements as it relies on secure enclaves and APIs managed by a central entity. While it abstracts private keys for ease of use, it does centrally control to some extent. And I'm just responding, Jesus Christ, because that was an actually useful reply

by an AI bot. So we have, if you look in the rearview mirror of like where we started this podcast 12, 13, 14 episodes ago, it was all about the AI agents. And the AI agent platforms like virtuals and arc and all these things. And everyone's like looking in the rear rear mirror as this like pile burning pile up of crashed agents because everyone has realized that they're all slot bots and they're not interesting. And I am reading Grock and Bankerbot be extremely informative in the replies and actually have like semi real connections to them. I think not everyone is as equally excited about this. I'll get into that in a second. I just tagged BankerBot and I said, yo, banker bot, what's in my wallet? I have not sent any,

money to this wallet. People have sent me tokens. But I've got I've got $11 of DRB. I've got $6 of veil. I've got $2.37 of banker. And this was like gave me a rundown and audit a list, a P&L list of all the tokens that I have in my banker wallet. I think it's pretty cool. Got like maybe $14 in there. Not everyone is happy about this. Base tweeted out this tweet that said in case you missed it, Brock launched a token on base. Thanks to banker bought and Klanker's trading mechanisms, Grock's AI-controlled wallet is already stacking fees. And I think people are just pattern matching between the AI slot-bought mess that's in their rear view mirror.

And so they're like, yo, that's not real. That's just like, you know, human-controlled, you know, puppeteering of AI bots. Sam Zizi son retweeted this tweet from base saying, I thought that was pretty funny. I'm begging the base team to have a little self-respect and not turn their chain into BSC, but with eagle noises. It's basically America's Binance smart chain, basically saying, hey, base, we already learned our lessons, let's not incentivize this like AI, these AI, these fake AI tokens. And I mean, if you go up and look at the, the deck screener, the decks of the DRB token, it just looks like a meme coin, it kind of is a meme coin.

But like, this is, this is basically running back Truth Terminal. We've seen this before. This is how Truth Terminal happened. It's the same, it's the same thing. It's a little bit different, though, in that these bots are actually useful. They're actually doing things. BankerBot is actually like. like a useful app.

And Grock is making useful replies. And I think it's kind of cool. I think it's a, it's not a zero to one evolution of what we saw with AI agents in the past, but it is much more refined. It's much more smooth. And the bots are actually kind of fun and interesting.

Yeah. Well, I would say V1, as you said, was truth terminal. And it would just talk to you and come up with like witty replies, David. But, and it would hallucinate a lot, right? Whereas this time, it's to your point,

a lot smoother, a lot more refined. It knows what it can and what it can't do. And most importantly, David, it can actually do things, right? You had BankaBot that was actually making purchases for people, you know? You had bankabot reporting what you had in a wallet that you technically owned. All he had to do was connect your kind of like Twitter account and authentify that way. I don't even have to connect my Twitter account. It already just mince me a wallet when I make it.

Yeah, yeah, that's right. I mean, if you wanted to like access it and like do some kind of like swaps or whatever. that might be. And I think the third thing that's really important to observe from here is it's still happening on X. It's not happening. It's not on Farcaster. It's not happening on Farcaster. Although, you know, technically these agents were birthed on Farcaster. By the way, to answer your first question why everyone has an MFer pick, it's because Clanker and Banker are like Farcaster-driven agents. That's kind of like where they... Yeah. And MFers are kind of like this Ethereum-aligned A lot of their community.

Farcaster aligned. Yeah. For the record, I have an MFer, and it's one of my favorite NFTs. I've got four of them. Yeah. Oh, yeah. Yeah, I forgot you had an MFER. I do.

I have a bunch. Yeah, yeah. Yeah. I have the only MFER that has both a propeller hat and 3D glasses. There's only one of those MFers, and that's the one that I got. That's my flex. That's my flex. That's pretty cool.

That's pretty cool. Do you like that more than your punk? Careful, he's watching you right now. He's right behind me, bro. He's looking at me. That smile's about to flip. Yeah, I have a more rare MFER than I do a punk, but my punk's pretty cool. He's got a nice smile.

I think a punk's pretty cool. 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-based, 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 Arbitrm 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. Introducing Unichain.

Built for DefyFi empowered by Uniswap, Unichain is the fast, decentralized layer two, designed to tackle blockchain speed and cost challenges. With this Mainnet 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 Eichane.

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 transparency and further decentralizing the chain. More than 80 apps are joining the unichane community, including Coinbase, Circle, Lido, Morpho, and Uniswop. You'll be able to bridge swap borrow, 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-exper 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. Visit Uniswap.org and swap on Unichane today. Yeah. So my takeaway with the whole banker, clanker, grok thing. Okay, I'm going to put my skeptics hat on for a bit, David. We haven't really made any net new improvement here. We have made marginal improvements. Yeah, marginal improvements.

It's still a bunch of people trying to game the AI to create a token or kind of. claim a token is theirs just so that they could pump it in a McDonald's pattern on your deck screen a chart. And I am yet to see anything different or otherwise here. But I totally agree. I totally agree. In addition to all of that, we just ran through like 35 minutes of trad AI capacity growth. And I want to connect those dots here because like we're talking about like there, no,

the agenic side of these models are becoming very good and very capable. And that is the new like focus point. like models are now, models are now highly intelligent. We're going to improve models, but they're already smart enough to do what we need them to do. Now it's about can we smooth out the clunkiness of the actual agentic side of these models and get them to do cool things. And so I think you can extrapolate the capacity of GROC in the future or other agents like GROC, Sontres GROC, but you could add on like more autonomy, more agency, more motivation into things like this. and all of a sudden these meme coins,

just like we saw the last time, can back into something cooler than just being a meme coin. Is the optimistic scenario? Yep. It's just since we're talking about markets and stuff and tokens on base. The banker coin actually like doing pretty well,

$34 million. And so this is like the, I don't know if it's fair to call it. I actually don't really understand the full integration between banker and clinker. I think it uses Klanko to an extent. Yeah. Yeah.

But how does banker, is banker just a meme coin too? they're all still kind of meme coins. Clinker collects fees. But I don't know what banker does. So I believe banker used to be originally like a product called TN 100X. Now it might be a ham coin or something.

I might be completely incorrect here. So don't take my word to be truth right now. But essentially what they helped do was you could tip people on Farcaster in whatever coin or meme coin that you had. And I think they kind of grew that into this defy abstraction layer, which is now banker, essentially. Right.

Yeah. Well, that's that. There's other AI agent, crypto AI agent news. We want to walk us through the kind of the recent news. For sure. Well, on the note, actually, of banker and clanker, Jeff put out this tweet late last week, which, you know, he kind of like tracks certain agent tokens every now and then and kind of like talks about, like, you know, what's performing well and what isn't.

This list touches upon a range of different projects, but I wanted to actually call out banker and clanker specifically. Because to your point, their prices have been like up only in a market which has been predominantly down only, particularly in the AI agent sector, right? Where there's a lot of fun on the macro level side of things and all that kind of stuff. But these are two projects that have been around for a while now with respect to like the agent meta. And what I found the most surprising with this, David, is they've been launching like three to 400 tokens daily, despite the entire meme coin meta kind of exploding in our faces, right? And this has led to its price 2xing since early February, despite everything else getting crushed. And it tells us that there is still demand for this service.

You know, whether people like pumped up fund or not, the fact is it has been the biggest money-making app in crypto this cycle, right? Also, it's worth noting that Clinker got its coin-based listing, which, like, somewhere in, like, the middle of February. And it spiked from $30 million all the way up to like $150 million. And it's since come down to $70 million. But $70 million is still like in the grand scheme of clinker token prices, a higher price than it has been. Yep, yeah. Great point. And then Banker, as you said, on the other hand, is a, you know, effectively a defy AI abstraction layer where you can kind of interact with it natively on X or Farcaster and it can buy sell swap or whatever for you.

And whilst it's not the perfect UX flow, it does make it much easier to go from discovering Alpha to then making the trade. Also, Banker has its own terminal too, kind of like its own Bloomberg terminal, where you can have a more personalized experience, right? But this isn't just random market movements either, right? If you pull up this other tweet, which shows where Banker sits in kind of like Defi AI Mindshare, this is a screenshot taken from Cookie.com, which is, you know, a platform that we look to to see market caps and mine share and stuff. Banker has been rating consistently at the top of this AI agent sector for a while now, David. I think it's been like three weeks right now. So just, you know, all of that to point out that, like, the insatiable demand to mint tokens is there now, even during a bear market, which is kind of really the phase that we're experiencing right now. It's pretty insane to see.

Now, typically, like, if you were to compare this, I guess the larger point is between these two projects is that they're both on base and they're outperforming the entire AI agent market. Typically, aside from the virtual ecosystem, base has always underperformed the Solana ecosystem agent projects. right. So to see this resurgence is a pleasant surprise. There has been an energy shift towards base at least in the last two to three weeks. Yes. Definitely with a call it the collapse of pump dot fund. Pumpd off fund is still making like quarter million dollars a day in fees. And so it's it but like it has come down from a very high amount and like pump dot fund. I don't think it graduated any tokens in the last couple weeks

to radio maybe just a few. So there has been like like pump like meme coin activity is down across the board, but maybe it like dropped in 50% on basis side and it dropped like 95% on Solano's side. Yep. Yep. Exactly. Yeah. Okay. So there's something in our agenda, EJaws, that confused me called model context protocol.

And I was like, what the fuck is this? And then you were like, bro, this is so important. This is exactly what we're trying to like talk about. This is the thing. This is the thing that anyone and everyone's been talking about in the real AI world and also the Web 3 world. Okay, so we're moving on. I'm pumped by this.

We're moving on from AI slop coins back to into AI fundamentals. What is model context protocol? Okay, firstly, before, I know everyone's really excited right now. It is not another blockchain. It is not an L1. It is not an L2. It doesn't have a token. Disclaimer, it does not have a token.

It's not a crypto project. Yeah, yeah. It's focusing on this one thing, which, you know, a lot of crypto people kind of avoid. It's called utility, David. And I'm super, super excited about this. Right. it's been all the rage in both the crypto AI or crypto in general and traditional AI worlds. And it's called the Model Context Protocol or MCP for short.

Instead of describing what it is, I'm going to start by describing the problem and then tell you how it solves it. So we have all these AI models and they're all super, super smart. But in order for them to reach their full potential, David, they need to be able to access and use things like, I don't know, websites, tools, databases. Slack, cloud services, commerce stores. I don't know. You get the idea, right?

And the problem is there's no easy way for them to do that. Like right now, the only way is by creating a custom integration for each application, for each model. So as you can imagine, that gets pretty expensive and time consuming and exhausting pretty quickly, right? So Anthropic, which is the creator of the leading AI model, Claude, came up with a bright idea. they thought, well, why don't we create a single software layer where all the apps, databases, websites, whatever in the world,

can just connect to once and then instantly be accessible to whatever AI models exist right now or get created in the future. So it's like this synonymous just layer. And that's the model context protocol. It's a single layer where if you connect your AI application, model, or whatever, you now get access to hundreds, thousands of databases and tools. Now, why is this important or why is it so exciting?

Well, now your models can stay up to date with the latest. So it's always contextually relevant. It's also able to do a bunch of stuff, talk about a bunch of stuff, relay information. Remember, chat GPT is cool, but only really as an information source. It needs to be able to, like, actually do things? Actually, David, it's now reminding me, do you remember why we got so excited about agent frameworks or why it's still such an exciting perspective? because they were able to not only allow you to build Web3 agents,

but also access a ton of different tools because the frameworks basically plugged into these different APIs. That's what this does, but on a unanimous layer across any different vertical Web2 products and Web3. Go ahead. To turn this into it, like, I'm still trying to wrap my head around this. Is this like a wormhole for context in the Web 2 world? So AI agents need to be plugged into more things.

And so we need this like middleware, knowledge, short-term memory, working memory, interoperability layer for agents so that they can access more parts of the internet inside of their understanding and context. Is that like that's going on? That's basically what it is. Yes. In nerd speak, David, that's exactly what that is. Not even nerd speak, just crypto speak. Actually, low IQ left curve crypto speak is what that is. That's literally.

That's literally it. And of course, I was initially skeptical about this release because Anthropic, of course, is a private company. It's funded by some of the top funds. So I was like, well, if they're going to centralize this, this is not going to be useful or it defeats the purpose. Well, it's not credible neutral. Of course. Well, it's actually 100% open source. So now you can add whatever.

So now this is like the polygon ag layer where guys, they're like, guys, this is totally credibly neutral. This is just our standards protocol. And like ZK Sync is like, I don't care that it's credibly neutral. It's got your brand on it. Exactly. And so we're somewhere in between there, right? Yep, yep, exactly. It's 100% open source.

You can add any kind of Web3 stuff to this. And in fact, you know, this can obviously level up Web3 agents massively, but we're starting to see Webthrough teams already integrate directly into this. Actually, I've got two tweets here for you where, you know, one of them shows Send AI, integrate an MCP server, which is a model context protocol server. It's basically, you know, its own custom integration. So now apps like Claude can have access to 30 plus Solana actions. It says 100, actually.

Oh, 100. Okay, so you're my little desktop agent that can use my computer, like I use my computer. Now can do 100 different things on Solana. There you go, David. That's roughly 98 more things than I've ever done on Solana. There you go. I have bought meme coins on Solana and I have sold. The fact is sold.

That's pretty insane. But yeah, it's cool to just, again, like, I always think back to this one episode you did with Ryan, this one time, David, like back in the previous cycle, where it was actually it was before the previous cycle, you know, those dark, dark days of the bear market where you and Ryan- crypto is cool. Yeah, yeah, yeah. You and Ryan on every single roll-up would say the following phrase, all of these things, David, is just it's laying the tender. it's laying the Tinder for the eventual fire that will burst into this big defy thing or this big

you know, NFT thing. And everyone thought you were completely nuts back then. And then we had defy cycle and NFT cycle and we were like, oh, well, what the hell? It's like all these different cool things to do. And I feel like this is the same thing happening except it's on steroids, David. It's on like some other type of wave

where these things are real, exist right now, and can be put into action right now? now. And in the Web 2 world, it's already being done. So with Web 3, I'm just waiting for like, you know, the cogs to fit together and for this machine to start wearing. It's going to get nuts. I think people right now, like, it's pretty desperate times in crypto, like Bitcoin. It's back over $80,000, but it hit $78,000. Solano was like down to 115. Ether is below 2K. So it's like really desperate times. And then people, people are just like looking for the light at the end of the tunnels. Like when, when are you going to have a new meta? Like, we need to. Like, we need to

need a new meta. The last meta we had was dog shit. And that was the meme coin crime meta. And before that, it was the high FDV low float meta. And then we had this micro AI slop bot bubble. And I think people are like looking for all right. Like what's the next meta in crypto? What's what's going to happen next? How are we going to, where's the next arena? Where's next casino? How are we going to go up in price? It's AI guys. It's going to be AI. It is like we're going to figure out a way because AI is the most capable thing. The first half of this episode was talking about how incredibly capable AI models and AI agents are becoming. And now the second half of the episode is like, and we still got these token things. We still got tokens. We're going to figure out how to slap these things together. It's

going to be another AI agent bubble. We'll call it a bubble. It's going to be better than the last one. And even the AI agent slop bot reply guy meta was actually not the first AI meta. That was just probably the first AI meta that you, listener, heard about. There was two more metas before that that were smaller. And so I think this is this is what's coming. This is going to be AI. Like AI is going to cause the next hype cycle in crypto probably. I mean, yeah, David. If we just pick up the Quito dashboard chart here, AI is still long and strong, right?

Right. 35%. That is the majority of that's green. Yep. Yep. And it's literally been like this is the lowest it's ever been over a yearly time frame. It's typically averaged around 50%, but it's never dipped below 30%, I believe, which just insane. And we had some brutal couple of weeks. Remember, no one's really talking about

the crypto AI thing anymore, David. Everyone's shitting on the slot box. Except for us. Except for us. We're still here. I'm going to be the new Ryan. Like, we're laying the Tinder for the next wave up and all this kind of stuff, right? But yeah, AI is still the number one mind share subject in crypto. Hands down. All right. Bankless Nation. All right, you jazz. That was a really good week. Man, there was a lot. There's a lot going on. There never seems like there's enough time to cover at all. But Eajas, thank you once again for helping us go through the meta. I'm bullish, man. Really bullish. So bullish. Yep, dude. Bankless Nation, you guys know the deal. Crypto is risky. Crypto AI is even riskier. That's because it's the frontier. It's where we want to be. And we are

glad you're with us on the frontier and with us on the bankless journey. Thanks a lot.