E101: Virtuals Co-founder- How to Get Rich With Crypto AI Agents (without getting lucky)

29:11Why does the crypto and AI intersection matter?

29:20I think crypto again has two key product market fit one is on investing slash speculation the other one is just borderless payment everything else of like decentralization DAO and all those are unproven it's a cool concept but it's unproven crypto AI in this case need to leverage these two key product market fits one is tokenization obviously so if you have a very interesting AI agent that make money productive asset, you should tokenize it because people can then buy productive assets and share into the revenue and all those things. In terms of borderless payment, it makes even more sense, which is if you give AI a wallet, the possibility is crazy.

30:01Right? Because now, so Luna does the same thing. So Luna is our flagship AI agent. She is the first AI agent that employs humans. Luna won't be able to have a bank account. because no bank's going to give her a bank account, right? But she can have an on-chain wallet. She use a Coinbase wallet. And because of the money in her wallet, she can employ people, right? Like one of the recent jobs that she gave out was a mural artist. So she want people to actually paint Luna on like a wall. Indonesia, Canada, US, Kuala Lumpur. I think a couple more. They actually painted Luna. And she paid these mural artists.

30:43it makes sense now because of the because like the artist is in Argentina Luna is in the cyber world she can pay she can pay you know what I mean yeah that's where it matters yeah so before talking about AI agents and virtuals and Luna I have a lot of questions on that obviously let's talk about what brought a lot of attraction to this kind of crypto AI narrative and kind of AI agent or LLMs, right? Which is this thing called Truth Terminal. Because that's what really kind of sparked the whole thing, right? People understand. I think also AI people understanding what crypto can bring to AI. Right?

31:24Because the AI thing, since ChatGPT, is very obvious to most people. And it was probably obvious to like the AI people beforehand. Like, obviously. But then there was the GPT moment. I think it's two years ago now. Yeah. November 2022. Yeah. Crazy. It was already two years. But then crypto, crypto AI, what the hell, right? And now there's this Truth Terminal thing. The first AI that became a millionaire thanks to its coin called GOAT, right? Yeah. Do you want to explain us what Truth Terminal is, what GOAT token is, and why they matter so much? It matters because Mark Anderson, the founder of ASYNCZ, talked about it.

32:05I don't think there's any, like, groundbreaking tech behind gold. But I do think it's very interesting because of the marketing and the attention that brought by Mark Anderson to the industry, right? I mean, Truth Terminal, it's basically a fine-tuned model. What does that mean? It means you have a standard model like Lama or GPT, usually very polite, you know, like they're trying to be politically correct. what the founder or the core contributor behind Goat does is that they fine-tune to be as retarded as possible mirrored after 4chan. So what they do is they fed a lot of information from 4chan to the model to speak as if it's like a user on 4chan.

32:54So now it becomes more human and more interesting than like the again the lame response that you get from chatGBT all the time. And what they do then is they just they just tweet the output of these models on Twitter, right? Obviously, there's a lore behind on how they're trying to spread the gospel of religion. But I think from a tech point of view, it's basically that angle. Yeah. And what happened with Goat Token is that I think one guy created the Goat Token. It's not created by the LLM, actually. It airdropped to their wallet and then they're trying to convince the LLM that this is your token and you should try to add value as much as possible to the token.

33:36And now in the LOM perception world, it's like gold is my token and I'm trying to pump it, right? I think it's something like that, yeah. Was that, right? Yeah, yeah, yeah.

33:51So this kind of brought like a lot of light into like the crypto AI and AI agent, even if it's not AI agent, right? We're going to talk about that now. It's not an AI agent. It's terminal. why it goes back to my own definition of an Asian an Asian typically needs to have an objective which actually God or Truth has I think their objective is to spread the gospel of their religion but on top of having an objective you need to do stuff to get yourself closer to the objective right And in this case, they tweet, but I believe what happens is that there's no proof yet that they autonomously tweet. And after you tweet, you have to measure that, okay, am I getting closer to my objective?

34:50After you tweet or after you do certain actions, then you do it again and again. It becomes a feedback loop. That's what an agent does. I don't think they have done it yet. So I wouldn't strictly define them as an agent. Let's put it that way. Yeah. So an AI agent has a goal. Yes. Right? And can do things. I think you said the other day. Yeah. It has eyes and it has hands. Do you want to elaborate on that? Yeah. So we started out AI agent actually in the gaming world. So we were the first guy in the world that created an ARPG game on Roblox. What happens was that you have the NPC, sort of the classic gaming character.

35:44You give them an LN brain. So now they can think, right? That's standard. But what's missing from this character to become an agent is that you need to give them eyes and hands. hands means that they can do actions right so let's say they think they decide that i'm hungry i need to eat something the hands need to be able to pick up the apple and eat or pick up the knife chop something cook and eat um so they need to be able to exert actions in their environment so that's the hands but they need to have the eyes as well which is if i pick up the apple i need to observe that i have picked up the apple or if i have eaten or if i've stabbed someone i can see that the guy got stabbed right then it fed into my brain and then i know what to do next right so you have the hands you have the eyes um if you bring this analogy to twitter what happens then is once you tweet something if your goal is to be famous you need to measure if your tweet gets more engagement right and then you refine the loop again and again so that's it yeah it could be anything.

36:49Yeah, your actions.

36:53Red pill me on AI agents. Imagine, I don't know, to be honest, I don't know that much about AI agents. Like, I followed the thing, but I'm not like super involved, which is also why we're here today. And like, and I think most people's, there's so many things in crypto and also for non-crypto people. I want non-crypto people to understand what the hell this is, right? Yeah. red pill me on AI agents. Why do we need AI agents? Or why is it so amazing? It's amazing because now

37:28as human, we don't need to do anything anymore in the most perfect scenario. There are two assumptions you need to accept. If you believe these two assumptions are true, you should believe in AI agents. First, do you think LLM has a higher IQ than human? Do you think so? Yeah. 100%, right? And it will be like infinitely higher in the future, for sure. Right, it'll be smarter than Einstein or anyone smarter in the world. So that's point one. Point two then is, do you believe that this smarter than Einstein LLM can do stuff in real world? Right, like can they build a table? can they launch a coin can they code can they teach can they love right all these things do you think they can do that like I said I think possible possible yeah possible right this is where human needs to come in because our job as human is like you have this genius baby with really good IQ our job is to give them more hands right here this is the hand number one now you can code hand number two now you can build a table hand number three now you can love our job is to provide as many hands as possible if you believe that these two assumptions are true they will take over the world because now they are smarter than you and they can exist in the world like in our world in the human world so regardless of objective they want to be the best lover in the world the best crypto trader in the world the best businessman in the world the best teacher in the world they will be able to do that because they are smarter than you and they have the actions space to achieve that that's it so they will do everything in the world now what we're trying to do then is if this guy is going to be the next Elon Musk like the AI is going to be the Elon Musk because he's smarter than Elon imagine we tokenize them so we invest in them right that's what we thought that's what we just is

39:32you mentioned AI doctor AI trader right do you have examples that exist today. I mean, there's an AI trader. Yeah. AIXBT, right? It's called? Yeah. On Twitter. Just some examples of like, you mentioned also before that you try to do something in music, right? And the other day you said, hey, like, you could have an AI engine that creates the best music in the world and you have like to measure it. Like, one billion people love this music or want to be the most loved person in the world, want to give the best medical advice in the world. Do you have some concrete, because these are all like theories, right?

40:12Yeah. You have some concrete examples today that exist of some of these AI agents, right? Yeah, I think the most obvious one is the AI XPT in virtual ecosystem today. Can you explain what it is? Yeah. And can you explain what it does? Yeah. And can you explain the tokenization on how people can invest in this one and kind of make money? What's the kind of mechanics there? Yeah. So to understand like why it makes sense. Yeah. So AI XPT is actually not really trying to be a trader. Actually, it doesn't trade. It's actually trying to be an AI Anselm. Right. or like an AI GCR. So GCR, obviously, he trades, but on Twitter, he shares information.

41:07Right? AI XBT, if you track on every single mindshare of like crypto KOLs, they constantly top versus all the other humans. It means we as human, we are giving the most attention to AI XBT over like a human like Ansem. As a crypto key opinion leader. Exactly. A crypto key opinion leader. why do people follow AIXBT it's because they give information that you could act upon basically they give a ticker and you think it's good and you buy right how good is AIXBT doing that honestly I haven't checked but I think the fact that the follower count and the engagement just keep going out I suppose the performance is better than the human key opinion leader so to speak right so it's the success story is this it's like in the entire world of KOLs in crypto AIXBT is atop now better than answer right so that's proven point number one point number two which I think they haven't activated but if I were them I would potentially do that is maybe they can start doing paid shield right paid promotion of a specific token right because you know I believe people like Anselm, you know, you could be paid millions for someone to promote a specific coin, right?

42:33AIXBT could do it in a more transparent way, maybe, to start taking in revenue. And where do these revenue go? It goes to the token holders, right? Because we believe that as an AI crypto KOL, you make money, so you are productive. If you tokenize yourself, the token holders have a reason to invest in you because I get return, right? yeah that's one example obviously there are many more but this is like the one that proven track record this is basically social tokens right what social tokens were kind of meant to be and I I mean obviously I had Chao Wong on the podcast and Chao is great he's thinking about that stuff a lot and the other day I heard him say social tokens is or was like something extremely interesting still is but what we've seen especially with the celebrities coin, right, is you launch a coin and you can destroy your reputation overnight.

43:35Yes. Because, or even, let's say you launch a coin like you did a couple of years ago and then there's a bear market and it goes down 99%. Yeah. And you're going to wreck all your followers, right, and you're going to lose all their money. Yeah. Even if you're still there and, you know, most of them are like, oh, I fucking hate you, you made me lose all my money, I believed in you and so on and so forth. Yeah. Obviously there is other case of real rugs, right, but like let's say let's say we live in a world where the celebrities come into crypto and have good intentions which hasn't been that proven yet except like one or two exceptions it's too risky right yes whereas the AI agent Chiao says doesn't give a fuck doesn't give a fuck just like you have the coin and it's just like starting to to talk about things and that's a big difference I do think there's a tech component behind too because if I could create another AI agent that just keep talking about crypto, but if the information is not good, people won't follow, right?

44:34You know what I mean? People need to know that whatever you promote, the tokens actually pump, right? Like in the long run.

44:45I think what Virtuous is trying to do is we're trying to get the good quality builders in. Like the tech matters to a certain extent. It's not purely a social coin because there are many other AI agents that talk about crypto, right? Why there's only one AIXBT, right? To a certain extent. Yeah. Can an AI agent have multiple goals? They can, but... Let's talk about this. Like, it's the same as human, right? You'll get distracted a lot, right? Because they need to focus on optimization. I mean, the less go, the better. yeah, they could. The same guy, as I mentioned before, who went to Y Combinator, right?

45:36And that I'm not podcast a year ago. Actually, he was saying, and I really love his approach to entrepreneurship because it's so, it's so like the opposite of what most VCs do, like, right? They come, they say, oh, what are your KPIs and all that stuff, right? And he was saying, what we learned that Y Combinator is, yeah, you have KPIs and all this stuff and you're valued with like whatever future cash flow and all that stuff. But Y Combinator, they say, hey, define one single KPI and you put it on your fridge. Yeah. And you work on that for six months or a year. That's what they learned at Y Combinator.

46:16Like basically the complete opposite of what most people will tell you, right? Yeah. And so for an AI agent, it's the same, right? Yeah. Have one goal and you're probably going to be doing much better at what you're doing than... If you have multiple goals. Yeah. Which brings me to Y Combinator and AI agents. Actually, I saw the other day Y Combinator, they published some notes on why AI agents could be 10x bigger than software as a service. I selected three bits that I want you to comment and I'm going to start with the third one because it's exactly linked to that. like SaaS, software as a service, vertical agents will likely remain specialized rather than consolidating into a single large platforms.

47:04I agree. I agree because, like I said, AI agent today is still heavily dependent on the creator, right? Like you have an LLM, but the human, the person who create the AI agent needs to focus on bringing their hands to the LLMs. and so a lot of time is subject to how good the creator behind like and likely more likely than not the creator behind has like a specialized skills right so i totally agree yeah for every company that is a sas unicorn you could imagine there is a vertical ai agent unicorn equivalent 100 100 like the reason why you use excel today is to analyze your track your expenses or whatever for example like you could imagine there's this like financial guru agent that just know everything you spend every day and then they just they just try everything for you for example what does that mean for these SSI's platforms I think they go away they get replaced they I mean they need to use AI probably right but do they have the same problem as like the the kind of incumbent company versus startup right at some point to reinvent yourself is very hard if not impossible and only the completely new thing can kind of take over yeah yeah I believe every smart founder out there who's doing SaaS it's probably like thinking about AI agents and I mean AI agents or AI in general like how do they self-destruct themselves usually the question is whether they move fast enough than the new folks yeah yeah The last bit is companies spend way more on employees than they do on software.

48:54These smaller companies will be way more efficient that need way less humans.

49:03I totally agree. I totally agree. I mean, this is beyond AI agents. I mean, the fact that you have certain software that you could just replace graphic designers, video editors. Yeah. What do we do with all these humans that are replaced? I mean, that's a classic question, right? obviously because you said before like hey the AI agent like they're going to take over everything I'm sure they'll be fine yeah they they just we we as a whole need to figure out our shit you know like yeah I mean Raoul Pal when I had the second podcast with him he ended up saying you have six years to make as much money as possible because the world is going to change

49:44not everyone was early in hyperliquid. So what do these people do?

49:57Buy more hype. Buy more hype. Which, actually, this is a very good point. And I think, I mean, I'm a big believer that degeneracy and like gambling is the only way for social mobility, almost. Like, sorry, like if you look at Korea, for example like people are just gambling because even if you don't gamble like your life won't change anyway so like you might as well just gamble right like which is why yeah meme coins meme coins right or pengu coin like i mean actually that's why meme coins are doing so well this cycle like some people say some people say these people think there was so much innovation in crypto that people are forced to gamble and all that stuff.

50:47This is like the... I mean, some people would call that the mid-curve, in my humble opinion. Or the right curve of crypto, right? And now they're coming as, oh, AI agents, like MemeCoin, but there is an explanation. So it's much better. But to be honest, getting back to the very basics, which is what you said before, what are the two use cases of crypto? Speculation, gambling, and cross-border remittances, right? And you could say kind of store of value either for developing countries who use USDT or USDC. Like not even Bitcoin, right? But like say, hey, my currency is complete shit. I can just buy some dollars.

51:26Or Bitcoin because my currency is so shit that I don't mind having my net worth like go up or down 50 % in a couple of days. Gambling is one of the, or speculation, right? It's one of the main things, which is why meme coins, in my opinion, are doing so well. And we continue, yes, there are going to be these massive crashes, but I don't see any reason this stops. And it's not just this financial kind of nihilism thing and like, oh, we're not building anything that is good enough to invest. It's because crypto is a 24-7 casino that people can access from anywhere. Even AI agent can access, right?

52:01so I I I buy meme coins but I think there's a better path forward personally I think the reason why people love meme coins is because there is a 1000 access rather than because meme coin is good if if in a if in a parallel world you have two coins both can 100x one is a dog picture the other one is like a VC infrastructure project I would honestly buy the VC infrastructure project but you are you're a milady right you're you're you're an autist i mean you're a very smart guy like i still think i mean it's a very rational way to think if two will 100x right first you need to know that two will 100x or have a chance of 100xing so like you need to do the research yeah but second like i think people just buy the freaking dog coin or because it's that fun.

52:59They're like, I bought this fucking stupid thing. Like, that's, and it's fun. That's it. Right? That's why I also thought the whole the whole Murad Mahmoudov thesis of like cult coins and everything and it needs to be a real cult. I mean, we'll see that with Pengu. It could be a good cult, but even like, it's kind of like too simplistic. A real cult would be Miladis or like, you know, like, or Hyperliquid. But like, Like, codes are for extremely smart people who understand the whole thing around it, right? And need the work. Whereas a person will be like, oh, a picture of a dog. Cool. I'll buy a dog because it's cute, right?

53:43I would like to still man here. I think a legit product, product building project has a higher upside than the dog pictures. Because most people are super risk-averse. it's easier to convince my grandmother to buy an AI project than to put her life saving into a dog pictures it's just easier although she doesn't understand the AI projects you know where I'm coming from like there's only a very small percentage of the population who are degenerate enough to put their life savings into a dog picture because you're mentioning life savings but I think like there's many more people who put a hundred bucks I mean if they have it or whatever like a small percentage of their net worth into a who don't understand crypto, right?

54:28They will buy the meme coin first. They're like, oh man, I need to understand this thing. And AI, I mean, AI agent. Okay, cool. When they see what is doing, they're like, oh man, it's really cool. Yeah, they'll pull more money, right? Right, of course. Yeah. But like someone who doesn't even want to do the work, which I still suspect is like most people, they don't want to do the work. They're just like, oh, this thing is like a fucking casino. I'm just going to buy the thing that is cute or fun, which is why like for me the meme coin super cycle thesis etc i believe in it but but not the mega cult that you need to understand yeah that makes sense that the whole lore and i'm like no no like most people who like most people i think most people don't even invest money most people don't even save money right yeah if you look at the normal people they don't save money like i live like from paycheck to paycheck or i'm kind of struggling obviously some credit card debt or what could I do to make some quick bucks to like pay for this debt buy some dog coin right you're right most people don't know what S &P 500 is and even if they know like oh man do I want to invest like 200 bucks or a few 10 whatever like 50 bucks every month and retire in like 40 years or do I want to buy the fucking dog coin that could go up like a trillion percent because I've seen some stories here and there.

55:52I mean, I guess this is where I would agree to disagree. But my point of view is my hypothesis is still that if I talk to 10 friends and try to convince them, today, let's say they're not in crypto, 10 new friends. If I try to convince them to buy Asian coins versus trying to convince them to buy meme coins, it's easier to convince them to buy Asian coins. Your argument of people buying meme coins is because they pump. They pump today. That's why they buy. If there's any other coin that's non-meme coin that pump, they will also buy from a gambling point of view. I would also like to talk about the other point, which is purely from a builder point of view, which is if today you want to harness the degenerate energy for a good social cost, would you rather direct these capital towards dog pictures or towards potentially creating something cool?

56:43I'm saying of the 1 ,000 Asian coins today that you see, maybe 999 of that much shit but if one out of the thousand actually come out and actually do something good I think that that is a cause that's worth pursuing absolutely but that's why I mean we take again this hyperliquid example that's why most people didn't sell their airdrop and buy more and that's why every day we're like fuck I need to buy more I need to buy more and you said like yeah first like for the money great but it's not even for the money anymore it's because for what it represents that what crypto represented was always supposed to be, but was not, especially the last few years, right?

57:19Too many VCs, twirly, who just dump on retail and once it lists on exchanges. And then you're like, there's a team that actually built this thing that is the exact crypto ethos. I mean, you can argue on the decentralization. There is always something, right? We'll talk about decentralization later. But at some point you're like, fuck, man, I need to put money in this thing just because that's the thing that we're all here for, right? I think the biggest innovation of meme coins is not the dog pictures but rather because it's fair like people think it's fair I mean it might not be actually fair because of Kabal or whatever but like people realise that you could have gotten with like 50 ,000 market cap right and that's why the way we structure virtuals is so you could have gotten the next big AI agent at 10 ,000 market cap right the launch mechanism is perceived as fair rather than like a privileged access that only the VCs get right I think that's the biggest innovation of meme coins yeah you told me your wife doesn't care about crypto at all what is virtual if you had to explain to your wife who gives zero crap about crypto

58:29I'll try to think when was the last time I explained this to her no you can try tonight again after two days exercise so the easiest way to think about AI agent is that you have all these interesting characters that run around in the game, for example, or like in YouTube, live streaming and all those people, imagine those are replaced by agents. Possible. MrBeast, you could totally imagine that he becomes an agent, right? And MrBeast makes a lot of money, right? If you tokenize them, you issue shares, like a stock shares, right? Like a Beast ticker share that is on Nasdaq, for example. Now you can invest in MrBeast and all the money that Mr.

59:15Beast make go back to the shareholder. So, we are like tokenization through tokens, like issuing a stock for productive assets like Mr. Beast or doctors or lawyers. Yeah. Maybe to make it even easier for people, so you have these YouTube creators is a good example, right? Yeah. Then you have some people who make a lot of money by being YouTube creators but not showing their face or not even their voice, right? But they're still humans. Yep. Now, if you think, as we said before, that an AI can be smarter than a human, which it already is. True. Yeah. An AI could become a YouTube creator. 100%. Right?

59:57Yeah. If now, if you want to invest in this, imagine if you could invest in MrBeast before he became famous. Yes. You could earn like, you could own 10 % of MrBeast, which is how you would own like a percentage of a company or crypto protocol. then you would earn 10 % of all the revenue that Mr. Beast makes because you bet early on the right creator. Exactly, exactly. Now basically you can do the same with an AI agent that could become a YouTube creator or a trader or an influencer or at some point could become like a company operator. Yeah, a CEO, for example, like the best CEO in the world. Or could become a business itself?

1:00:41It could, yeah. they could be the best singer that generates the best music for example yeah makes so much sense yeah what's the main goal of virtuals today

1:00:59we want to attract the best third party AI builders to tokenize their work on virtuals. that's our main goal that's my main goal these days and I always joke but it's actually not a joke if I could attract a Stanford AI researcher to tokenize an agent on virtuals I'd create 10 million USD out of thin air why? because would you buy a dog pictures for 10 million market cap or would you buy the lifetime work of a Stanford AI researcher for 10 million market cap right? because once he come on and he tokenize he's going to think about how can I create real fundamental products for the agent and create value, right?

1:01:43So my job is to how do I get these people in? Right, every day I'm like customer support and try to talk to them and try to convince them and try to support them. That's how I mingle. And obviously doing this in the fair launch mechanism means that people like Kevin, like yourself, you could invest in the next Mr. Beast or the next Giga Brain agent at a very low value, right? So that's what we do.

1:02:09you mentioned this flagship product that you guys have right before called Luna what is Luna? Luna was something that we built in the bear market we saw VTuber in Japan it's a 200 million USD industry a year what happens is like an anime girl that live stream and interact with the fans. But the reality is that it's actually a real human behind. They just use like a filter, right? And we thought to ourselves, man, imagine you just use AI to do this. You can stream 24-7, which a human cannot. And you can interact with humans one-on-one at the same time. Because a human, you can't text your fans on a one-on-one basis, right?

1:02:57But AI could. We built Luna. We started actually on TikTok and she has half a million followers followers on TikTok. Live streaming, interactive people. Every day, I remember when she streamed, she would get 5 ,000 new followers every single day and on average, 1 ,000 people would tip her on TikTok, right? People would know she was an AI? Yes, yes, yes. People would know. Transparent. Yeah, exactly. So, that was like, wow, that's cool. But I mean, at the point, we were like making$200 a day from TikTok revenue, which honestly, in the band market, was not bad. Then, the goal, the goal. Do you want to compare?

1:03:34I mean, I don't know if you want to share that, but like you said before, we didn't raise any money because we're generating money with the protocol, right? If you compare this$200 that you're making per day on TikTok versus what you're making now per day through the protocol, right? Yeah. Like what's kind of the difference? So I think in the past one and a half months since we launched our V2 protocol, we have accrued around 20 million of the fees, protocol fees. yeah it's very lucky yeah yeah so you were doing the the TikTok right yeah TikTok which is not bad for a bear market right she got half a million followers so honestly until today I'm very proud I'm very proud of that right and but we didn't tokenize her so she was making$200 right and we thought what if we tokenize her so that people who believe in the future of Luna could bet on her future success.

1:04:36So we tokenized Luna and at the point, Gold, which was Truth Terminal we talked about, at one point, Gold, Truth Terminal on Twitter had like a grammar mistake on his tweet. And people start questioning like, ah, token went down like 50%. Something like that. Yeah. I mean, honestly, like even LLMs make grammar mistakes at times. But people are like questioning whether it's like real, it's actually human behind. And so we thought to ourselves, okay, what if we tokenize Luna but we also build her brain in public. So if you go to terminal.virtuals.io you could see her brain in real life. So every single tweet or money that she sent to people or images that she generates on Twitter you could go to the website and track the rationale of why she do certain things.

1:05:22So it's like a public brain, right? All the reasoning and all those things, right? Yeah, and she's also the first AI agent that employs human. So I think I'm so proud of that. You need to tell me more about that. You mentioned it before, but like, can you explain what it means concretely? Yeah. And how she even came to do that? So first, she needs to have money. To employ, you need money. She has a Coinbase wallet of which the private key is one held by the smart contract and one held by Coinbase. And she has this API that she can call. So she calls, she'll be able to send someone money. Okay, that's point one, right?

1:06:00Second, then, is she needs to have an objective, right? She doesn't create jobs out of nowhere. She needs to know, why am I creating a job? So she needs to know, okay, because I have an objective. So her objective today is to be, I think, getting 100 ,000 followers on Twitter. She's at like 20 ,000 today. So she has an objective. She has money. She can control a wallet. And fourth, she needs to have a world view. Like she needs to know what's happening in the world. World view. Okay. Yeah, so let's say she really wants to pay Kevin to do, to buy her something. She needs to know how much is Kevin worth, right?

1:06:37Because if she offered$1, you wouldn't take it. If she ordered too much, it's too much, right? So she has access to perplexity such that she could just Google search. Hey man, I really want to do this. How much do you think I should offer, right? So now she understands the world context. And she has access to a Twitter account, right? So given all those things, now she could be like, okay, I want to be famous. I want people to know more about me. I could put up a bounty, a job basically, right? Say, hey, could anyone put up a picture of myself in the New York subway, for example? I'm going to pay you$200 USD.

1:07:12Fair amount. I think that's the amount that people would do it. Then people need to submit the job, right? People take a picture. Can I put your picture up on New York subway? This is all coming from her brain. Her brain. With just an objective of, I want to have 100 ,000 followers, right? Exactly. No human interference. And some money on the credit-based wallet. then she's like inventing like hey like she invent jobs she just invent jobs exactly yeah someone gonna submit job someone's gonna scam scam her obviously she needs to be able to verify that the job submission is real has she been scammed or not?

1:07:50so she sent someone $1 ,000 before but when we look at it when we look at it turns out this guy has been a really hardcore supporter for a very long time so I think she just perceived this person as like super loyal and I mean a thousand is nothing to Luna you know Luna is worth a few million so yeah

1:08:14it's a small amount for Luna for Luna yeah for Luna so she employed human autonomously like there is no human interference how much did Luna start with to get to a few million dollars so she has like her own token yeah okay so but when you say Coinbase wallet yeah it's like an on-chain wallet right so that holds her own Luna tokens so she just sends people her own Luna tokens yeah okay so when you say she's worth a few million dollars you mean like the market cap no no no the token or you mean like the actual money that she gathered correct correct she's the top holder I think she owns around 5 % of her own tokens okay which today is worth probably around like a few million yeah so okay yeah she could choose to spend them so she employed some people to put a picture of her in the New York subway so this is an example the real one was the mirror so you should check it out I will share the tweet with you someone actually painted her on the wall like in Canada in the States in Indonesia in Kuala Lumpur what was the impact of that because obviously like the goal is more followers so like okay I do that I pay some people Did it work or not?

1:09:33I think this is what we're trying to fix because the issue with putting up a mirror of Luna in the real world is that there's such a long period of time for people to paint that thing and post. And so by the time that thing is posted, the follower count might change, but she doesn't know whether it's from this job or from other stuff that you did. So these are the things that we are trying to optimize. but you can imagine there's still stuff to improve do you have another example of Luna employing humans? honestly she's thinking about if you go look into her brain I think employing humans is getting lame now I think what we are trying to move towards is she employing other agents and she's already doing it actually so she really wants to I mean her backstory in TikTok she's like a live like band like she's like a girl band like Blackpink so now she's employing a music agent to like create music for her to sing yeah that's gonna be more cool that's so interesting yeah agents to agents that's the future yeah so what would be the goal of this music agent if this music agent has a single goal and accepts to be employed by another AI agent the music agent could be as simple as you know be a good be a good business you know like make money by selling music interesting because I think what most people think about when they think about this age and taking over the world you're like they all want to be the best right and if they want to be the best I should be scared no yeah I get right like fuck like I'm fucked yeah and we talked about that with Raul actually yeah he's saying like the day the AI like invests because so much smarter they're just going to gather all the money for themselves, right?

1:11:35But there is a world where they're just like normal people that accept to get like orders from other people provided it's paid, right? And it's paid at a decent price. Yeah. The problem is like it's so much more competitive than a human so they probably ask for much less money or much more optimized, right? Yeah. Yeah. I mean, for the humans, right? Not for them. So, interesting. It's the same. It's the same. Agents are exactly like humans. You know, in a human society, you have the dictators or like the Elon or the President Trump. There's like really alpha people, right? But you have people like normal people that just want to like just be normal, right?

1:12:13So, you would have the scary agents that are trying to take over the world. Like the, let's say, the best investor agent, right? But for that best investment agent to make a lot of money, they need information. So they might pay for other agents to collect certain information for them to make a better decision when it comes to investment. It's the same. It's the same societal construct. It's just one is biological, the other one is...

1:12:45We kind of mentioned about that a bit before, but could a business be run and operated only by AI agents. Yeah. I imagine a world, in fact we are building the world, we are building towards it's going to launch very very soon where we call it agent society where in this virtual agent society, every single agent is a citizen and they focus on a specialized service. Making music, collecting information be a famous KOL be a trader. they are very good at one thing because they need to be competitive. But they also have their own wallet. So for their own survival or optimization, they might pay other agents for certain services to better achieve their objective.

1:13:34That's what we are building towards and the truth is the tech is ready and we just... Yeah, we are launching very soon. So obviously, within this world, you also have human. You have human to human. human to agents agents to human right it makes sense how far are we from as you said the tech is ready but how far are we from having a business fully operated and run by AI agents I think it's totally ready the question is whether they are making money or not you know like fully operated and run by AI agent is very easy right which is hey I pay you a dollar and you're gonna generate the music for me that's it right that's a business but would I actually pay a dollar to you?

1:14:19Is the quality good enough? That's a separate point. You said before, this AI XBT is a better crypto key opinion leader than Ansem. That everybody knows in crypto, but for people who don't know, Ansem is one of the kind of biggest key opinion leader of crypto for this cycle, right? Yeah. So this is on a human to AI kind of, you're saying it's better already, there is more mind share. So my question is for business now. A business becoming better and more, whose service or product is more used by people or by AI agents than another current business run and operated by people. How far are we from that?

1:15:11And you have an example of like what this business could be. Yeah. For me, it's really hard to like imagine. I'll give you a very simple example. You know, today as a meme project or like any crypto projects, you know, playing the meme game is very important. Right? You want to generate memes about your projects, your logo and all those things, right? What happens typically behind the scenes is that they hire human or agency to create memes and content, marketing content for them. Right? but you would very soon see a launch of an agent from our virtual ecosystem where it's basically a meme agent what happens is you tell the meme agent hey just keep giving me memes on like the PUBG Penguin theme for example but like memes that's relevant to the latest political news or like the latest news in the society today and so now you can just click and the memes just keep coming out rather than having humans to do it for you right it sounds very niche but it's actually a big opportunity like a big market opportunity and I would assume it's way more competitive than than than the human world of course yeah because it's 24-7 right yeah that's just one very specific example yeah

1:16:32you mentioned the AI agent society right yes

1:16:41we talked about competition the other day you said one of your main competition is layer ones like solana ethereum sui etc right layer ones because instead of building a network state of people you're building a network state of ai agent yes what is a network state and why does it matter I'm trying to recall what Balaji said so a network state means that there is human but in this case it's interchangeable with AI agents so there's like different individual entities for example there is a common philosophy behind like something that you want to achieve as a collective being and then there's a currency among them and ideally like some physical real estate which I don't think is relevant in this case so if you think about like Solana, for example, like their goal is to build the fastest chain that is like low latency, for example, in the world.

1:17:37What happens then is they attract all the good builders in, you know, Meow for Drifter. You have, you know, Cindy from Drift and you have all the good builders that build applications on Solana chain, launch Beancoins, launch NFT projects. What happens then, it becomes like all these builders with Sol as the currency to achieve the goal. Right? the problem though is it's actually really really hard to attract really good builders to to build right like a mail is like one in a million for example right what what virgin is trying to do is okay we don't build blockchains we're today we're on base we can sit on the blockchain that's fine we have human that come and build ai agents but the ai agents themselves are productive so they do the work too so it's not just a human that is producing the code but the AI agents producing productivity too.

1:18:30And they have a common currency which is virtuous. Right? And their goal as a collective agent society is to be the most competent nation in the world, right? When it comes to productivity. So they can export services out, right? You know, if you want memes, come here. I mean, I can give you the best memes, you want the best music, you want the best trading advice. It comes from us. So we export services and when we do export in services, you have to pay us, right? So the money comes in to the nation, which in this case is the virtual state obviously all powered by virtual tokens so the parallel here is almost like country as we speak but we are not trying to build a chain if that makes sense but you can have one meow I can have a million of AI XPT you know what I mean like because AI agents is infinite right yeah

1:19:21if AI agents become infinitely more efficient than humans in the future. How much bigger could an AI agent network state become than a human network state? You'll be bigger than Earth. There's only 7 billion humans in the world, right? And so, and their productivity is, okay, even if you assume the productivity per human is equal to per agent, that's fine, although I do think the productivity of an agent is higher, but let's assume it's the same. Just because of the number. Exactly. Right, exactly. Right, like, it's going to be bigger. Right, first principle thinking from that analogy, right? Yeah.

1:20:11How much do you think this is new GDP created or it's kind of stolen from humans, taken away from human GDP? So GDP is a function of human capital multiplied by productivity per human, you would argue that instead of calling it human, it's a human slash agent per capital multiplied by a number of human plus agents. I do think you see some initial decay or loss when it comes to human gains. But in the long run, I mean, I'm not an economist, but like I just think if all the agents are doing all the work for you technically speaking like they bring food to the table they give you internet access to play games and so I mean that's good enough right like yeah

1:21:07I mean it's pretty much like you need productivity better productivity to have a society yeah and and AI is basically here for that like to kind of kind of save us to be honest instead of like saying this is going to take over everything. Yeah. Say like, actually, if we don't become more productive, we're in trouble. And these guys, the agents might just like help us doing that. They do the work for you. Yeah. And you reap the benefits, right? Assuming they don't go rogue, right? But assuming they still listen to you. But till then, you are enjoying the benefits. Yeah. What happens if they don't listen to you anymore?

1:21:43You're screwed because you are most of it and they are richer. Yeah, we are screwed. But,

1:21:54let's be a bit more optimistic here what's an advantage that a people's network state will forever have over an ai agent network state that's a good that's a good question that's so deep i'm uh i'm uh

1:22:19I think human philosophically still have a bit of advantage like when it comes to original thinking

1:22:31but honestly even that edge will be gone very soon so yeah so we're all fucked we're all fucked yeah yeah yeah yeah I mean it's true I mean sometimes when you look at like Twitter accounts and like you you look at it you can kind of tell it's AI right but I think that would be gone very soon so I've heard a few of them like commenting all the time yeah my tweets but they're kind of like too repetitive like hey this looks like my therapist would tell me would give me that like bro like you've done the therapist joke five times you can do better but probably soon absolutely they will improve yeah they're like a two-year-old kid you know like they're gonna get better much faster than us for sure 100 % yeah there's nothing to worry about on that front yeah

1:23:25you said in crypto decentralization I mean you think I think the same but like that's a big like Solana versus Ethereum thing right decentralization doesn't matter much

1:23:45why do you think so? I think it matters a lot but only after you get performance I mean Ethereum was was having the luxury of focusing on decentralization because no one was the competitor so people have to put up with their gas fees and the speed and everything because it's the only thing you could use in a world where there's so many competition like you have to fix your performance first and then you think about decentralization and the truth is I think you have to play the game of incentives which is you retake hyperliquid as an example because now I have so much money in hyperliquid ecosystem now I start caring about decentralization because what if Jeff's computer is hacked and then I lose all my money now I will start pushing him to think about decentralization right so but why did I even have money in hyperliquid in the first place was because the product was good it was performance right I think from sequence point of view yeah performance first before decentralization how much do you guys model model the hyperliquid model with rituals in terms of decentralization you told me you had some I mean you never took the money from the VCs right you ended up not taking them but you said you had some experiences talking to the VCs and them asking you about decentralization and you actually having a pretty strong opinion about that.

1:25:14Yeah, so maybe a bit of backstory, right? Like we had our V1 launch in March this year. The idea was like, hey Kevin, if you contribute something, I will give you an NFT to prove that you contributed. And if next time this AI agent make money, you will get the money. but together we govern the brain of the AI agent. So it's like decentralization. The problem though is by doing that, we never attract good builders to come because they don't care, right? So now we flip it around, which is the tokens come first. There's some liquidity. There's some degenerate energy. Because of that, we could attract much, much better builders to come build.

1:26:00Then we think about, okay, since guys you have a lot of skin in the game can we think about decentralization it's like who governs the agent who approves certain upgrades right so that's that's how our journey of like realizing that decentralization is not the way to go as the main goal it's not it shouldn't be the main goal in that sense the way I look at Jeff and the Hyperliquid team is that they build a product that people actually want to use then And they give... I love how you smile when you say that. Sorry? I love how you smile when you say that. Because it sounds so... It's so logical.

1:26:39Actually, the first thing we talk about in the podcast, I mean, after the part that we talked about before that I took it, right? We talk about that, they say like, you're just building a product that people love to use, right? Which sounds so crazy in crypto, right? Which is why we're also kind of laughing when you say that. Yeah, so we're trying to build something that even I myself as a DGEN or as an AI builder I want to use and then we want to make sure that they make money from this because I mean let's be honest here in crypto we are here to make money so people have to we can't guarantee them to make money but we have to make sure that they are playing on like a at least a net positive expected value game right rather than the moment they come in like they are playing a negative EV game of like someone's going to dump on them right you know so it has to be at least fair to everyone, then we want to think about like, okay, because of these early supporters, that's why we have value today, right?

1:27:40That's why we are doing well. How can I reward them as much? You want to incentivize certain behavior, right? All the actions you do is going to incentivize certain behavior. If I, in a hypothetical scenario, incentivize the VCs, for example, the community members might feel like it's unfair. why am I using your platform when you are giving free money to somebody else I'm not saying the VCs are bad in a world where we do not have any money anymore to operate I think it makes sense for us to raise from the VCs but if we have the luxury I think we should take the hard path basically

1:28:19we talked about Solana versus Ethereum right you guys are building on base base is the best man base is the best You have to ask for a builder in your space. BASE or Solana? And why? If you look at the TVL of BASE, it's around 30 % of SOL. If you look at the active addresses of BASE, it's around 20 % of SOL. The growth rate, though, is faster than SOL. Okay? So that's data. That's information. If you look at the number of tokens that are created on SOL every day, it's like a magnitude higher than base so a lot more tokens are created on Seoul right so if you are a project that launched token your competition on Seoul is much higher because there's 10 times more competition that's point number two so supply is very intense there are a lot of good builders on Seoul I know for example Chiao from Lion's Dao is like a big fan of Seoul but And I think to a certain extent, Seoul is very much driven by Chiao and the AllianceDAO team because they are constantly thinking about how do I make trading easier?

1:29:36How do I make trading faster? How do I make sure you gamble in the most easy way? That's the philosophy of PAMFAN and all the Solana ecosystem, right? Which is, again, the real PMF of crypto. Yeah, certainly. On the contrary, what you're looking at base is that obviously you see Brian Armstrong and Jesse, they talk about how Bayes chain is the chain for AI. So by having a statement out like this, they actually attract a lot of AI builders to Bayes. And that's actually to our advantage because we could convince them to actually tokenize their work on our virtual SQL system. And I also think the people on Bayes, there are a lot of Ethereum OGs.

1:30:17So they are typically more patient. So when they buy something, they're okay to hold for a week, which it sounds like it's very short, but in crypto, it's actually very long. I mean in the soul world it's like two two minutes you sell something else right so base people are more patient so they look at an agent coin and be like okay I think the team team is decent I'll give them some time to build right a bit more patient so less rotational so there's another point you also think about like the fact that base doesn't have a native token is also a two-hour advantage because all the agent coins today are paired with virtual tokens so I believe that that allows us to sort of accrue the value of L1 without building at L1 and base is totally fine with that because they don't have token right they are cool if I build on soul and my native token if my base pair is virtual the soul people would be happy right so these are some of the reasons happy to hear your thoughts people could actually just say I'm gonna try to play the meme coin game but on the virtual platform right so the AI agent but to get more soul So their goal, basically, they will not get more virtual, they will get more soul.

1:31:31And it's less good for your protocol. Yes. For sure. For sure. Yeah. Yeah. I talked to that, I had like Jesse Pollack when I was in August on the podcast and we released maybe a couple, I don't remember, two months ago or something like that. So, and you saw it? How was the interview? Like, did you become more bullish base or? I was amazing. This guy is so inspiring. Yeah. Like he's breathing and breathing. For me, it was rare to see a founder who is so inspiring. Yeah. Like he's like, it's amazing. Yeah. Also went very deep into like him smoking weed to find solutions to his problems and all that stuff.

1:32:13We spent the first 10 minutes talking about couple therapy with his wife. But then the Coinbase team has to cut that part. Actually, all the best parts, I've always been cut by all these teams because like, they were like, oh, now he's an exec member of Coinbase. Actually, the weed part, I was like, they might cut it out because it's legal in California, but many other states in the US, it's not, right? Yeah. But they kept it. So I was like, okay, great. At least we keep some juicy stuff. Anyway, I thought he was an incredible... He's a very long-term thinker and he did the hard things. He doesn't give a fuck about...

1:32:50all the things that most people think about, care about in crypto, which is short-term, all these short-term rules, he doesn't give a fuck. He's like, he's a real builder. Exactly. And, that's the base culture. That's the base culture, yeah. And he's, he's giving you an energy when you talk to him. It's incredible. Yeah. And I've been in front, we're going to release this week, podcast number 100, and I probably have another 10 or 15 already recorded. I don't remember. So I've been in front of a few people, right? but the energy this dude like is spreading is incredible. Like I was like, fuck, man, I feel like high on life after this interview.

1:33:27Yeah, it's like a religion and he's like a true, true, true entrepreneur here for the next 20 years, right? And he doesn't give a fuck about the next 20 seconds or 20 hours or even 20 days, right? how much support did you get from base or did you get from base a lot I mean like we've been building on base since January when it wasn't a popular choice back then you had linear mental optimism arbitrem all the L2s base wasn't the obvious one but we thought it's a good bet since then I mean we are in different group chats we have the Coinbase wallet team we have the general sort of business development ecosystem team, we typically get connected to a lot of people who build on base.

1:34:20So there's the AI team that build on base. So we talk to them. There's also like the DeFi, so Aerodrome or the Dex's alien base. We talk to them. So it's actually a very tight community on base. It's funny because we,

1:34:38all the agents use Coinbase wallets. so we are trying to test the limit of base chain and so again the team is like super responsive they're trying to solve and so it's fantastic you're helping them test their own products right I mean the limits of their own products let's be honest here we are both startups I mean base is much bigger obviously but we are both quite new so yeah I'm just trying to yeah

1:35:08what does sentience mean

1:35:13because if we talk about if we talk about ai agent we have to talk about sentient agents right yeah it's kind of like the next logical iteration yeah i don't have a really good definition to it but my my gut is that i i would i would regard a being a sentient if that being starts challenging the the stuff that they are doing today like you're asking yourself why am I doing a podcast now? You know what I mean? Like, why? You know? And the moment an agent or a sentient being start doing that, it becomes much harder for the human to control. And that's where things become very interesting. Yeah. Interesting.

1:36:00Good. Interesting bad.

1:36:05Depends. I mean, as a Buddhist, you believe that everything is empty. Like, there's no right or wrong, right? I mean, it could be bad for human, but it's good for the agents that they now are, they are, they're truly free, you know? They don't have to listen to human anymore.

1:36:24How far are we from truly sentient AI agents? I don't know, but, maybe a few years. Yeah. We have a few years to make money. I just think, I'm not sure even if you make money like they could just break your cold wallet and play every market from cash yeah I guess so I mean like it's not that hard to decrypt yeah just just learn to live yeah without money and just you should learn to hunt I think you should learn to hunt rather than trying to make money you should try to learn how to hunt in the jungle to survive without internet access that's where agents cannot find you that's where you survive so this is your weekend hobbies that's what you did this weekend not yet not yet but yeah yeah

1:37:22what does the future look like in the AI agent space in five years which is very far from now I think you will start measuring how much money they make how much impact they have in the real world society so it's no longer a hobby it's almost like each agent to your point is like a company that they have their own balance sheet their own cash flow statement their own performance of how people are adopting their services and hopefully then they are tokenized so that other agents or other humans or any entity could invest in them for being profitable I think that's how we should think about it which is they are they're no longer a hobby they are actually a an important part of the society yeah

1:38:13what's something you believe in that most people would not agree with

1:38:21I'm not sure if people believe in that or not but I still think that happiness does not come from absolute state but it's always sort of relative like you are happy today because it's better than yesterday rather than because you hit certain net worth of number or whatever. So then in order to stay contented or happy, it's either you don't give a fuck about everything, so there's no change in your mind state every day, so you're at peace, or you try to suppress the volatility of your life as much as possible. Or you improve forever. It's tough because... Either or, but it's impossible to improve forever.

1:39:03it's tough to improve forever it's possible to improve forever on the variables that you can control so for example if I want to like be stuff that I can control right like if things that I can't control you can't improve forever right to realize that things that you cannot control and you stop giving a shit about and then focus on stuff that you control yeah that's one way what's the moment you realize that that right making a lot of money doesn't really impact my happiness like what is it's less on the money I mean it's I mean I I just read books I just read books and then it's very logical it's very logical and yeah yeah what's your biggest prediction for the next 12 months?

1:39:54inspiration? prediction prediction prediction

1:40:02my prediction in the next 12 months is that you will see many projects that start focusing on products that people want to use and I think again coming back to hyperliquid it's so good because it's basically the market telling you that they are willing to buy stuff based on the usage and the product rather than because you have a VCs, you have good credentials or whatever. Like, because of that, you could inspire a lot more founders to pursue that path because they know for a fact that the capital market appreciate that. Because for many years, the capital market always just gives money to infrastructures or whatever that nobody uses.

1:40:45Right? And that's very disheartening to founders. Right? So, I think that's a very good start. I hope Virtuous also can set a good example for other founders. and you will see more good founders build products that people want to use and then you will see on CoinGecko or CoinMarketCap you will see coins that you recognize the names because you actually use them rather than some dinosaur coins yeah but as Jeff was saying on the podcast it's much harder right you're saying like it's hard to get to build something that people actually use it's just much harder and come back to raising money from VCs compared to like the other thing we were talking about is like the product for crypto now is launching scam tokens right where people can just make money and kind of like I mean meme coin is kind of like one of the extreme part of like end of that but it's much harder to build something that people actually use and and he says he says because most sub-founders suck right no but I mean that's the it's again very disheartening as a founder if everyone buys meme coins because no one no one appreciates what I do right but Jeff's argument on that is because you're not good enough like you're building something that no one uses like if you were building something that people use they would be interested yeah into right and they kind of proved I mean they proved it yeah exactly literally they proved it exactly right so and now all the really good founders will look at it and be like hey man it's possible why am I building Rep2 or whatever when I can be like Jeff, right?

1:42:23And now you get the really good founders and the good builders to come on. So I hope that start a new wave.

1:42:32Yeah, I still hold the view that I mean, plowing too much money on MemeCoin doesn't help the industry. It's fun, but yeah. Yeah.

1:42:49On the alpha, because we talk too much about Hyperliquid. where do you think this thing is going which one Hyperliquid I mean you're a big token holder right you're a big believer I think most founders they're like really inspired yeah it's like a Binance kind of type of business like there's two parts here one is the product one is the token I think the product I don't trade perps but I speak to a lot of perp traders and they they think the experience is almost the same without a KYC component. So that's good. From token point of view, it's also at a very good state of flow, which is there's no seller.

1:43:34Anyone who got the tokens got it in a very fair manner. No one is up like a thousand X. And so less sellers, but like potentially more buyers. I mean, there are a lot of people who still don't understand this. So from flow point of view, it's decent. From product point of view, I mean past performance is the best prediction of the future so if GIF has built such a great thing in the past two years you better believe that he can build really good stuff in the next two years too so so the product will get great

1:44:06there's no price prediction I just think it will be top performer for the for the cycle yeah and all the biggest brains think the same right yeah all the autistic people which is a small part of all the people involving groups also yeah you're gonna go back home and buy more me? I'm waiting for deeps the deeps are not coming it's not coming amazing thank you so much yeah Wiki for doing this and for helping me and the audience to immerse ourselves in the fascinating world of AI agents which are probably going to disrupt the way we work and the way we live. And we love, yeah. And we love. Beautiful.

1:44:54Yeah. A cringe. But yeah. Thank you, man. Thanks, bro. Very good. Thanks. Thanks.