Loading summary
Transcript98 lines
- [00:00]
A
Hey, it's Sterling K. Brown from the Hulu original series Paradise. The next chapter begins as Xavier's search for his wife takes him above ground. And what he finds will change everything. It was never just about the bunker. Tune in and discover the truth lies outside Paradise. The official podcast is now streaming and stream paradise on Hulu and Hulu on Disney.
- [00:23]
B
Even relationships can be faced by agents. One of the biggest questions I've been posing to some of my friends was like, could there be even a scenario in a world where an entire males or females. Right. They are actually replaced by robots? Could humans still survive in the era? Right?
- [00:44]
C
Hey, everyone. Welcome to Unchained, your no hype resource for all things crypto. I'm your host, Laura Shin. Before we get started, a quick reminder. Nothing you hear in Unchained is investment advice. The show is for informational and entertainment purposes only, and my guest and I may hold assets discussed on the show. For more disclosures, visit Unchained Crypto.com did
- [01:05]
D
you know that figure is giving away $25,000 in USDC? They're a decentralized digital asset platform for earning, borrowing and lending.
- [01:15]
C
Download the figure markets app using our
- [01:17]
D
link FigureMarkets Co Unchained Deposit into their democratized prime pools and earn about 9% APY paid hourly while you enter, every dollar you keep in for 25 consecutive
- [01:33]
C
days counts as an entry.
- [01:37]
D
Again, the link is FigureMarkets Co UnchainedDP. For full details. If crypto taxes feel overwhelming, you are not alone. That's why Crypto Tax Girl, a team that's been helping crypto investors since 2017, is offering $100 off on one on one crypto tax help. To get $100 off your crypto tax services, go to cryptotaxgirl.com Unchained. Again, that's cryptotaxgirl.com unchained.
- [02:07]
C
Today's guest is Jansen Tang, co founder of Virtual's Protocol. Welcome, Jansen.
- [02:12]
B
Hello. Pleasure to be here.
- [02:15]
C
Yeah, I'm really excited to chat with you. And part of it is because this is just a crazy time in the world of AI agents. So it's actually great that we're talk know. Today, Virtuals has an announcement that it is launching EastWorld Labs, an AI accelerator to foster the deployment of humanoid robots and to create a hybrid society of robots, virtual agents and humans. Tell us about eastworld Labs.
- [02:43]
B
Yeah, so I think for us at Virtuals, the we've always believed that AI agents are going to be fully autonomous economic actives. And that has been the grounding Principle in whatever we we've done across the year and a half or two years, we've built a few things, right? We've built one. The one that we're probably most famous for is the capital formation layer for these autonomous agents. People call it a launchpad, gives agents transaction fees whenever their token gets traded. That's one. The second one that we then went live half a year ago was about building a stripe or an agentic commerce protocol for this autonomous agent to work with other autonomous agents to produce economic value greater than the sum of its parts. And that is where we've realized that our whole construct or working with agents has been really in just one domain we've been working on with just digital agents, right? And we realized that there's a massive, or I would say a dimensional level of unlock in things that we can do with agents when we start involving physical AI, that is robotics. So we've been actually started to invest in robotics teams about a year ago, but we started dabbling harder ourselves about three months ago when we started working with the whole data layer for robotics, trying to understand like where should we play. And then we realized that in the end our cards lie in building an ecosystem. And we've proven that we've probably built the largest agent ecosystem at the intersection of crypto and AI. And we've realized that we should double down on that strength. And that's why we've realized that maybe what we should be doing in the robotics sector is to build a vertically integrated lab. And the idea behind this lab is threefold. The I. But we have one very core goal. And the goal here is that can we make it easy for founders who wants to build humanoid use cases at a consumer surface area to have an easier life, make it easy for them to one, raise funds, two, garner the data and the models needed to train these AIs and three, obtain distribution at scale. So that's where we started to realize like the best way to fit this all together is some form of incubator or lab or makerspace. And that's where we started toying with this idea of eastworld where think of it as a location. It's a co location where great founders, builders from both the academia side as well as people who are more on the GTM on the BD front to come together and hack stuff and start building things that is going to be interesting and that maybe some of them will also tokenize and then give the degens some assets to play with as well. So that's the core That's a core goal of IsWorld that we are launching.
- [06:01]
C
And it will it be like other accelerators that, you know, have different cohorts at different times, different seasons. So there will be, you know, applications and then you'll run the program for, you know, I don't know how long each one will be. So there'll be classes.
- [06:16]
B
We run a very unique mechanism. So it's going to be. There's two ways to get into this residency, quote, unquote. So the first is we tied it very closely to the asset issuance platform that we have. So the idea here is that if you're a founder and you are keen to experiment robotics, most people will start off with an idea, a white paper, a pitch, right? Do the same thing. And what we encourage people to do is then launch a token around it. So if you can. Because what we believe is that when projects tokenize, the markets become a very strong indicative factor on how well they will receive your idea. That's the way to beat your token up. And in this case, we say that if a robotics token that is launched crosses a market cap of $5 million, we then offer this guy a seat in the residency. And this seat in the residency includes pretty much a free humanoid robot for them to test the fleet of co workers that we have or technical people that we have at their disposal, and also access to some of the university collaborations that we are working with as well. So we've also tapped into folks at cmu, NTU from Singapore, Shanghai, Jiaodong and Oxford initially right now to actually push some of these training of the autonomous policies. And these are the kind of benefits that these founders can tap into when they tokenize. So that is path number one. Path number two is for founders who probably are not keen on tokenization, but they have a great team, great idea. We also are personally incubating teams. So we write a check initially and then we will get up a seat and hack together with them. So there's two paths to actually join iswu. One is the on chain tokenization path, and the other one is through a. Think of it as like a seed funding path.
- [08:16]
C
Okay. And do these teams get together like in a physical location to work together?
- [08:22]
B
Yes. So we've actually opened a 10,000 square foot space in the heart of KL and that's where actually we are moving our own office as well too. And so you'll be working pretty much with the virtual team. And then we have this robotics lab here in KL and alongside this incubator for People who. Malaysia. Malaysia, yeah. And we have, we have so far 30 humanoid robots that has joined the fleet. So it's probably one of the largest robotics testing facilities in Southeast Asia. If.
- [09:04]
C
Okay.
- [09:04]
B
And by the way, the count of people who bought these robots. Yeah.
- [09:08]
C
Wait, I'm sorry, say that again.
- [09:10]
B
Because we spoke with the team, with many of the guys who sell humanoid robots. Right. Unitree themselves and the other providers and we realized that actually the fleet that we've got is actually one of the largest fleets in the world.
- [09:26]
C
Oh wow.
- [09:27]
B
In one co located space.
- [09:30]
C
Okay, so tell me a little bit about what type of applicants or projects you are looking for for the accelerator.
- [09:41]
B
So we've. So again, it's open, right? We have a. I would say learning that, you know, sometimes we just don't know what good looks like. So, you know, we just have to take in any kind of ideas. But the one, there's actually one hypothesis that I personally have been pushing on and we've been funding this idea with our incubator, which is along the lines of remote teleoperation first before full automation. And we found the market basically think of it as business process outsourcing. You know how the call centers that you have today that generally a lot of companies in the US or in a more expensive country that would outsource their call centers into the Philippines, into India. And the reason is because of the rich arbitrage, it's cheaper to get a guy do the phone call to handle the phone calls in Philippines. The question that we had was what if it's also cheaper if you can outsource physical labor? And imagine this, right? Imagine a robot that is deployed in the States, deployed in Australia, deployed in UK that is remotely controlled by a human in a cheaper country like Malaysia, Philippines, Vietnam. And this robot does a lot of the labor that you can actually find this same skilled laborer in Malaysia. Think of it as like security guards to retail sales managers to the guys in the hotel to go and clean your bed to even slightly more complex stuff like plumbing mechanics, H Vac cleaners. And realized that we did the maths and the cost savings for even as of today's economics, it's about 40 to 60%. And the reality is the price of this robotic stuff is going to just drop over the next couple of years, right. So we've realized that there's going to be this initial gap in the market where teleoperation can start making commercial use cases as of today. And then that is apart from just cost savings, these use cases can also Generate a lot of real world data. Because when a human does all these like work in a very chaotic real world, that data of all the robotics joints and whatnot becomes very valuable for training fully autonomous policies. And the idea is then each of these teams who are setting out to solve different use cases, they over time will start training their own fully autonomous robots. So then over time they will self replace their own business model of a human teleoperation to then fully autonomous operations. So what we've been then doing is we've been funding use cases at the AT like this. If you draw a two by two is a. On the one axis is how much wage arbitrage you can capture. So places where, let's say for example a plumber in Australia is paid like $180,000, but a plumber in Malaysia is only paid like $10,000. Right. So that's a. In Saint Bridge arbitrage. So that's one axis. On the other axis is the need for a humanoid form factor. So again like you know, in a factory for example, that where every boxes are equally squared, you actually do not need a humanoid. It's not the best form factor to serve that use case. But let's say plumbing for example, plumbing everyone's house pipes are built different. The shape of a toilet is built for a human to actually access it. So this is where those places where a humanoid form factor actually would drive. So we've identified like 10 to 20 use cases already where in the near term it makes sense and we've been funding this, give you a few examples to be very concrete. Right. Think of it as like retail sales, restaurants, the guys who do the back office stuff in a hospital, the people who go at autoclave stuff carry all these equipments around. H vac technicians, plumbing mechanics. So these are a few areas that we've identified that is that, that we see a lot of value capture with this tele. With this like initial human. Sorry, robotics, ppo.
- [14:22]
C
Okay, wow. So something that I need to understand. So you said that you have this fleet of 30 humanoid robots. Are they all the same or are they like different? Like are they specialized in some way?
- [14:38]
B
So they, we. They're all the same. They are, they're the, they're the max spec unit 3G1s which are pretty much the top of the best robots in the market today because they're the most dexterous humanoid robots. Yeah.
- [14:54]
C
Okay. Okay.
- [14:56]
B
They can pretty much backflip. They have hands that can, that can grab things, hold things.
- [15:02]
C
Okay. And I guess like the other Thing is, so in some ways you're. You're talking about, like, I. I just want to understand the teleoperation. So you're saying that maybe there's a robot that works, like, you know, let's say here. Let's like, let's say like in the New York area, but it's controlled by somebody in Asia. Is that what you. Okay, yes. And so basically, they're like, replacing a plumber. They go around to different places and they'll fix your toilet.
- [15:35]
B
Correct?
- [15:36]
C
Oh, okay. So I guess, like, yeah, tell me. Tell me a little bit about, you know, how you think that will work. So let's say I'm. I'm a person living here in the US And I. Where. Where am I going to find this, you know, robot plumber?
- [15:58]
D
And then how.
- [15:58]
C
Yeah, tell. Tell me about the whole.
- [16:00]
B
There's gonna be a. There's gonna be a. There's gonna be a few. There's gonna be a few, I would say, forms of how this might play out. Again, this is very early, so things might change, Right. The one of them would actually be. Think of it as every household might even just purchase their own humanoid robot to store at a house. And then the idea is that this humanoid robot might have different users, Right. Most of the time, you probably just need it as a mate. So you have a mate that's actually just teleoperating, controlling the robot to clean your house and just do the basic mate stuff. And then one day you realize that you have a broken pipe, right? Something gets stuck in your toilet, and you need an actual plumber to go and do it. So instead of calling a plumber, you just need to switch the operator. So the soul of the robot gets replaced, Right. So a plumber comes in instead into the robot, and then he then fixes pipe. If you need to go and cook something at home, a chef enters the soul of the robot. So it could be one robot with multiple souls that. Different teleoperators that come in, or it could. It could just be a service, right? Like, David, you need to call the robot to get a robot delivered by taxi to your house, and the robot comes and fix the house. So. So, I mean, there's a few ways you can see this thing operating.
- [17:17]
C
Yeah. I mean, honestly, the first one where it's like, I have my own personal robot, and then they do multiple different things. For me, that feels more like adopted.
- [17:29]
B
Easy to adopt.
- [17:30]
C
Yeah. Instinctual to me.
- [17:32]
B
Yeah.
- [17:32]
C
But then it's like, then they're. They're not making money for Anybody. It's just I have paid them and they're saving me money maybe, is how to think about it. So, so like, yeah, then I, I guess it's a little bit because like, you know, where you started is you have these virtual agents and then they are out in the world like doing things and they're engaging in commerce. Whereas yeah, this first version is more just like, I'm just a consumer and I'm buying. Yes, okay. Okay.
- [18:11]
B
It will take time, but the idea is that if you take a step back from a big picture perspective today, this is the extent of what these robots can do. Right. And that's why the teleoperation is honestly, it's not the sexiest story, but it's the, is the work that will get us the largest amount of data to start building autonomous robots. So that, that will be a year, that would be a year and a half from now. Right. And the idea is then you actually start needing no humans, probably just one human supervising a fleet of 20 or 30. And most of the work is actually done fully autonomously a year and a half from now. And I think that's when the, each robot can start potentially be its own economic actor. And that will be the dream. Right. And honestly that's why we named the lab eastworld. It's actually a call out to Westworld, the HBO show. I'm not sure if you watch that show.
- [19:10]
C
No, I haven't.
- [19:12]
B
It's a theme park. It's basically a theme park where the NPCs or the non playable characters in the park are actually intelligent humanoids. They look like humans and then the humans go in and then they live a life in that story with robots. Right. And that's the idea. The idea is that there will be a world where these robots will become autonomous and they own their own economic sovereignty. But we need to lay down the foundational steps to get that. So these are the, whatever we spoke about was like step one, step two, Right.
- [19:46]
C
Okay, so can you tell me what those robots would cost? Because you know, there's probably a calculation that people could make about, you know, whether that's going to like if they were to purchase one, whether that would actually save them money or just be very expensive.
- [20:03]
B
Yeah, so, so the like the max spec G1 today costs about 60, $60,000. This is a maximum spec. So the, the, but the ones that you actually can, can tweak the hardware and then they have like fully automous like dexterous hands, the cheapest G1s in the market, I think $20,000. So it's a, it's a, it's, it's becoming more and more affordable. But initially, honestly I think the, the first adopters of this because even $20,000 is still a lot for a household. So we, we, we will likely see a lot more POCs with probably commercial folks first because they get a lot of more, I would say, marketing firepower. So a lot of the initial POCs that we've secured have been with retail sales, like stuff like hotels and stuff like that. Because to them, apart from just saving costs, when I have a robot in my restaurant or in my hotel, it looks cool, it's going to attract more customers, so they will need to pay more for that. So I think that's, that's probably where you see most of the adoption layer first rather than like household retail people.
- [21:17]
C
So I'm sure you heard or, or maybe not, I don't know. But shoot. There's been so much political unrest in the U.S. i'm blanking on when this happened. But there, there were some protests somewhere at some point in the not too distant past where there were these Waymo cars that people were, you know, setting on fire and doing all kinds of things to. And I just wonder, you know, in the case of these robots, like, especially in the case of like a plumber that's going around, if it's being remotely operated by somebody in a foreign country, even a different continent, you know, what happens if you know, somebody does something to damage the robot or incapacitate it? Like how do you deal with that kind of thing?
- [22:10]
B
Yeah, so I think it's a, it's an evolving thing. We'll probably have to learn along, along the way. But the idea is you probably would definitely have a team on the ground. You will have a operation team on the ground that will have to manage the fleet because I mean, robots will break, the parts will catch fire. They will have like a screw that you need to fix. So you definitely need an operation team on the ground. And over time the question is how more and more autonomous they can be from be it the point of deployment to the point of where the executable work. So yeah, but yeah, I know, I foresee in the next year there'll always be some form of human oversight, be the guy who actually just fetches these robots around or some form of surveillance system. But the beauty about these mechanisms even as of today, right. Is actually you probably can see those videos online where robots that can actually defend themselves. As a human, you probably don't Want to find a robot? I mean, this is quite funny because when we got our first shipment in to our office, this was like a few months ago, four months ago, and we were just testing it in the Wii work and we didn't know how, like how. What to expect of it. It was the first time like really working with a G1 and then we did like some controls that made it roll and he did some kung fu moves and the space was so limited that we all had to jump away from it. And then he started smashing into like our computers and he was like flipping tables. Like it felt like, you know, like he could just stand up and the entire table would just flip behind him. Right. So that kind of strength in their servo mottos is pretty powerful. So I don't think people want to mess with them if they know that these rewards can actually react back.
- [24:04]
C
Yeah, okay, interesting. Well, one other thing that I wondered about was, you know, just hearing you describe that they could replace all these different types of human workers. Like, don't you think there could be a backlash? Like, yeah, I just wonder if you've thought about.
- [24:26]
B
I mean, I mean, I think it's probably the. It's not a problem for robotics. It's not just a problem for robotics. Right. It's a problem for, we see with AI agents as well. I mean with the whole claw. Sorry, not claw, the whole Claude. 4.4.5, 4.6 opus. The GPT Codex. Right. You've seen that. I mean we've seen themselves even in a company that some of the junior engineers job becomes quite redundant. You'll be like, like a product manager can actually do the work instead of needing and a plus a senior engineer. You do not need your junior desk anymore and you start seeing jobs getting potentially replaced. And that's really for white collar workers. Right. So I think there would definitely be a whiplash. But it's not going to be just directed at humanoids or robots alone to be directed against the whole AI industry. Right. Yeah, but I think that's a question for governments to solve. Right. Sometimes I think a lot of times we can't do it. We can't solve everything. I think I'm on the camp of like, you know, we just bring innovation and we speed it up and the world will just adapt to these innovations. Right.
- [25:39]
C
All right, so in a moment we're going to talk a little bit more about, you know, what these robots can do and how they're going to work. But first a quick word from the sponsors who make this show.
- [25:47]
D
Possible. Want a chance to win $25,000 in USDC figure? A platform to earn yields, borrow against
- [25:56]
C
crypto and access lending markets is running
- [25:59]
D
a $25,000 USDC sweepstakes tied to their democratized prime product. Here's how it works. Download the Figure Markets app using our link FigureMarkets Co UnchainedDP deposit into a democratized prime lending pool and leave your funds there for 25 consecutive days. Every dollar equals one entry. So $1,000 equals 1,000 chances. While your funds stay in the pool, you're also earning around 9% APY paid out hourly. To learn more and enter, go to FigureMarkets Co UnchainedDP, which is also available in the show Notes. If you're looking for help with crypto taxes, Crypto Tax girl is offering $100 off for Unchained listeners. They provide personalized crypto tax reports and returns, and spots before April 15th are limited/ Unchained to save $100. Once again, the link is cryptotaxgirl.com Unchained.
- [27:05]
C
Back to my conversation with Jansen. Actually, before we move on, I did want to ask one more piece on that kind of, like, crime aspect, which is, could there be security issues with the robots? Like, what if somebody remotely directs their robots to perform a crime? Like, would it be easy to trace it back to the human that, you know, was controlling it?
- [27:29]
B
I. I would think so. But I mean, there's two parts, right? One is, can the fleet get hijacked? I mean, the idea is that we build proper systems in place for that not to happen. But in the reality of, like, people controlling robots and performing crimes, it's actually possible. I mean, like, yeah, I mean, in fact, in fact, it's. No, in fact, actually, you can make it even. Not traceable. Oh, no, actually, no, I know how they can trace it because, I mean, you can hide your. You can hide all your digital tracks, right? I think that's quite easy. But the part that you cannot hide, it's how was. How. How did the robot got imported into the country, right? Who was the. Who was that or who. Who. Who gave these bad actors access to the rewards? So it's slightly traceable, but it's going to be a whole, like, robotic cybercrime unit coming out in the FBI.
- [28:27]
C
Yeah, yeah, we're laughing now, but who knows? In five years, okay?
- [28:33]
B
That's why you need. You need to have these innovations that happen, right? Because now if everyone has a reward, their robot security guard can protect them against another robot's crimes, right? So it's like an antivirus and a virus.
- [28:48]
C
Well, that's assuming everybody has a robot. So anyway. Okay, okay. So you know, as you talked about, like, you have these robots, but, you know, some of them will get specialized. So how are you training them? Talk about how they'll kind of like learn their tasks.
- [29:07]
B
Yeah. So today we are collecting data in two forms, which is actually one of the verticals of eastfo, which is this massive data lake. Right. And the idea here is one, we've actually created incentive systems for people to submit. We call it egocentric data. Think of it as, imagine if you wear a meta Ray Ban glasses or hang an iPhone around your neck and that is capturing videos of the hands, of what your hands are doing. You could be folding clothes, you be cooking some food, you could be like making a watch, whatever that might be. Right. So that data is being collected. That's one type of data. The second type of data, which is while we are performing all these teleoperations, you get data of the human with the visual and also all the information of how the robot is actually moving all the joint data. So it's how each of the hands turn the amount of force they're placing on the object. So that is the second form of data, which we call it teleoperations data. So these two data sets, we preheat houses in a lake, right? Because there'll be different teams that we are working with, every team that we fund, and everything that goes through East World as a whole will basically start contributing data to this lake. And the beauty is then we hopefully become one of the largest robotics data sets for teleorbit egocentric hopefully in the world. And then this becomes also potentially a reason for more and more residents to come in, because then they get access to this data layer to then train their autonomous policies and they go to market with them.
- [31:02]
C
Okay, so, you know, now, now let's talk about kind of how this is going to combine with everything that virtuals has already been building. Because obviously you have the tokenized AI agents, the virtual ones. You have, you know, ways for them to like fund themselves on chain. So talk a little bit about what tokenization adds to the physical robots and. Or vice versa.
- [31:29]
B
Yeah, so there's actually, honestly in my thesis is that there's only two reasons for robots and crypto to actually exist. And the intersection of these is one, crypto acts as a tokenization layer, I. E. If a team wants to build a project that tokenize, they get attention. The attention they get around a token benefits them in two ways. One is that they get direct economic value because when people trade their tokens, they get a tax out from it. That can be quite significant. We've seen projects raising like seven figures in funding from just this taxation mechanism. So that, that is one reason, right? And the second, sorry, and part of this reason as well is the attention flywheel, because when, when holders hold your token, they become advocates for your project as well, right? They might actually be your users that you sell, your first 10 robots that might actually buy it because they hold a big bag of your tokens. They want you to succeed. So I think that that has been the, the clearest crypto PMF and is the, that's what we've seen also for a lot of digital agent projects. The second bit, which is probably not something that's here yet, but it will come, which is basically blockchain as that coordination layer for robots. Think of robots, again, as like code, right? They're machines. And for machines to work with machines, the best way is to use smart contracts, right? And it's pretty much what smart contracts are designed for, right? And this is what we'll probably see where if I'm asking my robot to work for me, and if my robot is in a restaurant being a chef and he's cooking stuff, right? And then he needs to deliver that food to customers out there, he needs to call a fleet of drones or sidewalk robots, right? That commerce or the interaction or coordination will likely be best performed through the blockchain or through smart contracts. So I think that's probably the second reason for robotics and crypto to coexist. But for now, I think for us, we will focus a lot more on the first part tokenization portion. I think that's get the kind of fly views of attention and capital for founders, and then hopefully more and more of these will become more autonomous. And then we start seeing the smart contracts part of crypto becoming adopted by robots.
- [34:07]
C
And then one thing I wanted to understand is so when they're making money, it does all of that revenue go to the owner of that robot? Okay, so there's no value accrual to like, like the virtuals, token holders in any fashion?
- [34:27]
B
No. So I mean, there's two parts, right? Yeah, virtuals take a small piece of their transaction. So when you charge about 1% of trading taxes on every token that is issued from the platform, the Devs, they own 70% of that 1%. And virtual as a ecosystem collects 30%. So 30Bs of every trading activity fits the virtual treasury so that's, that's that that benefits virtual ecosystem as a whole. We use that money to grow the, the ecosystem. But there's another benefit from, for virtual holders, which is every project today that launch in the ecosystem is an airdrop. About 3 to 5% of the token supply will be airdropped to people who hold stake virtual tokens. So think of it as like if I'm a virtual holder and I stake my tokens, I perpetually get every asset that's being issued on the virtual platform.
- [35:33]
C
Okay. Okay. So, you know, we have talked a lot about the robots, we've talked a little bit about how it interacts with virtuals, but this is actually the first time you are on the show and that we're talking about virtuals. So why don't you tell us, kind of fill in the backdrop of everything that's happened with virtuals since you burst onto the scene back in 2014 during those token terminal days. And you've developed a number of things along the way. Some of these things being mentioned, like the Agent Commerce Protocol, which powers the economy. There's also Butler, the platform for humans to collaborate with their agents, Unicorn, which is the launchpad. You know, some of the technical developments are you added the X402 payments protocol for listeners who missed it. We did an interview on that in the fall. It was super interesting. And then recently you launched this virtual revenue network. And I think, you know, part of that had to do with the opaquad thing. So yeah, just talk a little bit about, you know, the, the different developments and really what has been like most surprising to you or most interesting as you've been watching your creation out in the world.
- [36:46]
B
Yeah. So I think it all started, it all started when Truth Terminal and Goat had that run in 2024. Right. And then there was a spelling error by the AI. Not sure if you remember then. And then the Goat token crashed like 80% because everyone was like, oh, now there's actually a human. Right. It's not a true agent. And we've been working with autonomous influencers. We've been building autonomous agents for gaming back then. And we realized that that was like our way to show the world that actually truly autonomous agents can exist. So then that's where we actually entered the market. Right. We brought Luna in, put her on Twitter and people saw her terminal of her brain, that people were like, oh, okay, this is what a truly autonomous agent looks like as an ex influencer, like a Kol Twitter Kol. And then we gave her control of A crypto wallet. I think we're probably the first in the world where an autonomous agent controlled a crypto wallet. And I think people got quite mind blown because they realized that if an AI can control money, they can exit influence. And that's where, that's where things started flying, right? All people got inspired. They built their own versions of autonomous agents. AI XPT came to the scene and then the launchpad started flourishing. So I think that was step one. Right. Step two is then we realized that we had so many of these agent builders, agents and the diversity of different types of agents and we realized that we started to see specializations happening. Some agents were focused on like being a strong crypto kol. Some agents were focusing on, on chain execution. Some agents were focusing on like yield farming. Some agents were doing other stuff, right? Entertainment, healthcare and whatnot. Right. And then we realized that what if these agents can then work together? So we tried to bring them together and we realized that there was a ton of issues that agents face where they just tried to work together. And when we were brainstorming, one of the guys came out and said that hey, why don't we utilize smart contracts? Isn't smart contracts designed for no human in the loop? Trustless execution? Right? Code can trust code. Then we realized, yeah, actually it makes sense. So that's where we started coming up with the paper for the agent Commerce protocol in February last year. This was way before people like Google did the whole A to A. They published their standard like two months later. So we were probably the earliest in the industry. But then, so across the year we've been basically building different use cases. We've been trying to build different, I would say minimal viable clusters of agents working together. So we did that in like June, July, August. And then we then realized that one of the biggest barriers to agentic adoption is how do you make it so easy for a human like me and you to actually use and AI, right? Because a lot of times a lot of these usage are all hidden behind like complex, even like today, even open cloud for example. Right. It's a complicated thing, right? To set up your own Mac Mini or your own cloud instance to spin up the agent to spin up a terminal. Like not many people know how to do that. And we then realized that if we create a interface on X to let humans feel what agents can working together can look like. And that's why we launched Butler. So Butler was basically a chat. It's basically an account or X that you can actually chat with that can be personalized to you and it can help you do things, right. You can literally ask him like hey, I want to trade crypto's volatility, right? Can you help me? Long, long volatility this weekend. And then the butler would then go and figure out, right he would go into this agent commerce protocol. He'll find different agents to help him, information agents, agents that can trade options, agents that get structure options and then they build out the whole portfolio for you without you lifting more than your finger. So that's the idea behind Butler. So that was to improve adoption, right? We're trying to get more demand side, more revenue coming into the commerce protocol. And then recently with the whole open crawl thing popping out, we then realized that this is the entire market here that we've been waiting for. Basically we wanted more and more people to have agents in their back door. Because when you have an agent working for you or, or being your personal butler, an agentic marketplace, that makes sense, right? Because your agent would need a method to perform commerce with other agents. Could be to buy your stuff, to trade for you to, I don't know, to find you the news or whatever it might be, right? That's where the agent to agent network will start flourishing. So we realized that we had to write the, the adoption hype. And that's why recently we've launched AGDP IO which is effectively trying to position our ACP as the Amazon for personal agents. Because now the idea is like okay, if you are a developer, you build a service agent. If you are a user like myself, you spin up your claw, you just install with just one line of code and you have now access to a, to hundreds of different agents that can do different things for you. So we've also put in, we basically did a bit of an all in. We put in our entire protocol revenue into this initiative. We said that every dollar that we make as a protocol, we just incentivize growth on this Amazon marketplace or ecp. And that's what we did. So like last week, so it's a week of epoch. We've, I think the price worth of today is about $300,000. So it's all the every single dime that we've made as, as a protocol and we've just reinvested back into telling people that hey, if you are a top seller, you get a bonus incentives. If you are buyers who are testing different products, different agentic products, you also get a piece of this incentive pool. So again, all of these is meant to build up towards our North Star Right. Our North Star is agentic gdp and we believe that, yeah, agents are going to work and be economic powerhouses. How do we start showing the world that crypto matters? Crypto is probably the only way for these agents to interface at scale without any information loss. And yeah, and then the next thing that we're going to do is that to bring them into the physical world, but just always will be okay.
- [44:13]
C
So you know, as, as you were mentioning, you've been watching this openclaw mania that's been taking off and I wondered, you know, I saw you announce plans to integrate OpenClaw with ACP. So I guess, like to me that says you don't necessarily view it as competitive. Like how do you think about openclaw?
- [44:35]
B
It's a. Because think of it as like, I think the best analogy is we are building a stripe. OpenClore is building Shopify and Stripe will become a billion dollar. Only became a billion dollar company because Shopify existed. Shopify allowed everyone to build an E commerce website. So E commerce activity started skyrocketing. Stripe became that payment layer for E commerce. I think of it that way. Right. Or the economic layer for E commerce. And that's why Stripe did well. So for us, stuff like OpenCloud is valuable, very valuable, because it puts agents in the hands of more people. Hence there'll be more buyers and also more sellers on acp.
- [45:22]
C
Okay, so essentially you're kind of building more like the Rails. They're building something where it'll be more the users. Yeah, I mean, just generally it feels like you're offering what is not only a very competitive space, but there is kind of like a big race, you know, in, in so many ways. So I just also wanted to ask even about things like, you know, as far as I understand, virtuals has its own identity layer or reputation layer. And of course, you know, Ethereum now is adopting ERC8004 or they just launched it. So, you know, how do you even think about that? And I know maybe that, you know, I'm not sure exactly. I, I guess it's like not, you know, what's the word? Directly monetizable. But it is something where, you know, how much people use that system will affect kind of how popular, you know, whichever agents, you know, use the direct or, or maybe agents can use both. I don't, I don't know.
- [46:24]
B
So actually I can give clarity. So in fact, so when we built out acp, we were thinking of actually just covering the entire end to end because back then there was no player or provider right into the market. Then folks like X402 came out as a potential payment mechanism for agents or code. And then Ethereum came up with the 8204 and in fact what we've done is we've actually integrated them. I think one thing I want to put a note out there is that SAPP is built honestly the goal to be a public good. Right now you actually can fork the contracts that we've built and use it. And we are actually in work to try to work with some bigger foundations to really push for open sourcing the entire agent commerce protocol and letting Dask contribute to it as well to grow. So it's again designed as a public good. And for you can see like when X4.0 came out we were one of the earliest to support and to integrate them I. E. We let agents use X 4.02 to pay for the jobs created on ACP. And if you look at X402 scan today, I think probably we are the largest facilitator on x402 and if you then, and then when 804 came out we were also there to support them. We actually work very closely with them until today. And in fact we, we, we uploaded our entire registry of agents to populate their, their registry which I believe is going to be good, right? I mean it's, we work in a very open way. I think for us where the value capture comes from it's mainly from being a, A, a largest two sided market because with all these infrastructures coming up, even integrating them, ACP as a whole is going to be a public good, right? And what we then build is think of it as like a facilitator or like a, a infrastructure on top of ACP that we then take a small cut. We say that okay, if you want to use Butler for example, it's convenient because we sponsor your guests, we make it easy for you if you want to be a seller on our facilitator, we sponsor the gas for all your wallets and your transactions. We've built in a couple of extra value add like escrow agents, agents that can actually insure your transactions, stuff like that. So all this value add that we bring in will allow us to command a bit of a fee structure on top of this ECP infrastructure and that's where we get the value accrual. But honestly the way we look at it is the more people built in this sector is going to be good. I think mentioned this a couple of times before on many interviews Where I think back in the days when we first came out, there were a lot of competitors in the space, like folks like AI60Z, a bunch of other launchpads. I'm not sure you guys remember Eliza OS and whatnot. Right. And my answer to people was that those were in fact direct competitors. Right. These are infrastructure builders, so we tend to just build together. Those who are direct compete for attention, folks. And my answer was that the more competitors are in the space, it just means that we're doing the right thing. The most dangerous thing that we are doing is we are building something alone and only we care about the problem. It's a problem that no one cares about, and that's going to be more dangerous. Right. It's better to have like, you know, couple of competitors, couple of sharks in the ocean is how, you know, there's actually fish in the water. So, yeah, that's. That's how it stands.
- [50:04]
D
Okay.
- [50:05]
C
So then the next thing that I was wondering is, you know, you also have this token, the virtuals token, and I wondered if you could talk a little bit about how you think these different. Either, you know, products that you have, you know, as part of the virtual zone community or ecosystem, and also like the wider world of agentic activity, like openclaw, like all these things, how does that benefit or not benefit the virtuals token? Like, what do you see as the direct, you know, things that happen in this world that will benefit virtuals and which ones don't necessarily have a relation to the token?
- [50:46]
B
Yeah. So I think two parts. Right. And the first part is, I mean, the understanding of how token price moves sometimes is actually quite disconnected from fundamental motions. It's sometimes about narratives. I mean, I mean, that's. I'm super honest here. I. I've been a. I've been a crypto trader myself for like, I mean, way before I started putting. Since 2016. Right. And the reality is, as long as we are consistently innovating, we're constantly at the front of the curve. I think people will value us that we are always either the creator of narratives or the earliest within a narrative. So that's where capital likes to bet. So that's on the design side right now, more the. More like hard stuff. The way we design virtuals as a whole, we'd like to look at ourselves less of a application or project, but more of like a network state or like a nation state in that sense. Right. It's the same thinking like the way
- [51:50]
C
Biology Srinivasan would, in a sense.
- [51:52]
B
Okay, exactly right. I mean, one of the inspirations of eastworld is to build it as a network state for AI builders and robotics. We're actually working closely as well with the network school to run like builder hackathons and stuff to find out with.
- [52:09]
C
Oh, okay. Yeah, because he's based out like, I forget, somewhere near you, south of Malaysia.
- [52:15]
B
So it's like five hours by car away.
- [52:18]
C
Yeah, okay.
- [52:20]
B
Yeah, so, but so when we look at it as like a nation state, there's a few things, right? So the first thing it's if virtuals is that, you know, the base currency for any kind of economic activity in this ecosystem. And this is where the biggest, I would say tokenomics value add has happened. So whenever a project launches within the ecosystem, they have to pair with virtual token. And what it means is that if more and more projects come in, there will be more and more liquidity pools that has virtual pet with them. If there are a few of these tokens that become unicorns, it will basically drive up the value of virtual tokens as well. Think of it as like, you know, Korean stock market, right? Like if you want to buy all the RAM companies, Samsung and whatnot, you have to buy the Korean dollar first before you, the Korean one before you, you buy the stock. That's what the Korean one will appreciate in value. So it's same thinking, right? As of today, about 5, 6% of total supply of virtual is actually locked in agentic liquidity pools. It's quite, I mean, it's a decent amount. Our goal is to grow it to 25%. So that will just create natural supply crunch, hence more pleasant price action right on the token. So that's one, right? And then the second thing is we look at it as a, again as a nation state, you charge taxes for activities happening in the nation, right? When ACB is happening, there's a tax that is running and those taxes are then kept in the treasury and it's going to be redeployed for growth, to get more citizens in, to get more agents in, more robots in. Right? So that's how we think of the value capture of virtual tokens.
- [54:13]
C
Okay, So I did want to ask one other thing that's kind of more just, what's the word like inside the plumbing of virtuals. But I'm sure you took notice of the fact that base has announced plans to create its own stack move off the op stack. Virtuals is primarily on base. But I did see you've also added Solana. I don't know if maybe there are other Chains. But I wondered if the base plans were going to, you know, affect virtuals in any way. Like are you going to plan to stick around on base or you know, generally how do you think that move will affect you?
- [54:53]
B
So I think we've, we've been one of the first projects Bas honestly like we've, we've helped base general activity. We've benefited a lot also from BAS and Jesse and the ecosystem. So I mean we don't see ourselves moving away from base anytime soon. But there is a big question and I actually was speaking to Jesse about this like two days ago and we were. Because think of it, all of the agent commerce bid, they're all happening on chain and on good days it's fine, right? It's very cheap. But there are days where when ethereum price falls like 10%, gas would just spike across every L2 and then suddenly our network stalls because gas becomes so prohibitively expensive that agents just cannot trade with each other. So I spoke to Jesse about this. So we need to figure out what is the solution. We've also explored a R with a very interesting team who basically builds. Think of it as like blockchain but not sequential. Because the problem blockchains today, they are all sequential, right? Because like you know, every transaction has come after the next because you try to build a ledger. But for applications maybe the sequential mechanism is not the best. So we've actually run an R and D pilot with a team doing that. But there's no concrete plans to either build a chain yet. So we still try to assess as of today base to fit. The idea is hopefully the basic can even compress that call data and the efficiency of the chain and that it's going to be better.
- [56:41]
C
So yeah, wait, I guess I don't understand the part about not having it be sequential because if it's not sequential, doesn't that create the opportunity for like financial shenanigans or exploits?
- [56:53]
B
Yeah, very true. But if you think of. So if an agent, if one agent and agent A and agent B they're working together, that connection should be sequential. It's like we agree to do the job, there's an agreement put in place, you put the payment up, there's escrow, there's a delivery of the work and there's the release of the escrow, for example. So that should be sequential. But if there's agent A and B working together, there's agents C and D working together, there's agent E and F working together, nothing to do with each other. Right? But for them to work with each other, they can actually be in parallel if their jobs do not interface at all with each other. So I think that's where we realize that if you want to scale this using smart contracts and using blockchain to service agentic commerce, you need some form of parallelization in place.
- [57:49]
C
Okay, yeah, I mean, I guess then also you need to make it so that nobody can transact with, you know, Agent A and Agent G or, you know, whatever.
- [57:59]
B
Yeah, something along the lines.
- [58:01]
C
Yeah, yeah, yeah. Okay. So, you know, this has been just so interesting to hear you talk about, about all this stuff. And I kind of want to go back to something, you know, that we touched on a little bit at the beginning, which was, you know, why you built this accelerator. And I just wanted to hear you flesh out like what your vision is for, you know, when, when. So obviously you're trying to foster this society of like robots, virtual agents, humans. So tell me, like, what does that look like? You know, when this comes to fruition, what does our everyday life look like? You know, I'll wake up and you know, I'll what, have a robot in my house or. Yeah, just talk us through a day in the life.
- [58:49]
B
I think the idea is that in fact, even relationships can be replaced by agents digestion. One of the biggest questions I've been posing to some of my friends was like, could there be even a scenario in a world where an entire, I would say either males or females, they are actually replaced by robots? Could humans still survive in that era? So I mean, these kind of questions that we even discuss or postulate on, but I think for me, where do I see the world going is as simple as every stuff that you don't want to do, there's a agent or a robot that's there to do for, to do that work for you. If you don't feel like working, there's a robot that's going to make money for you. If you don't feel like cleaning your house, there's some something doing it for you, right. If you don't feel like talking to, I don't know, your wife going to get a robot to replace it. Talk to your wife. So my point is, yeah, I don't
- [59:56]
C
know if the wife is going to like that.
- [59:58]
B
If you do it well, you know, it looks like you feels like you. It's like Iron Man. Was it Iron Man? There's a few shows they do there, but people start doing stuff that they want and that matters to them and everything else becomes commoditized. Free and probably even below your conscious memory. Right. You don't even think about it. So, yeah, I think that's probably where we are heading towards. But I think one thing we've also realized is that there will be a collapse of the current economic models, like stuff that we know today. We've been building a world and economy, tourists around scarcity, and we're going to live in a world of like, I would say like insane productivity slash abundance. So, yeah, things will be very weird. Like whatever we see today versus like 15 years down the road, it's going to be very different. Like, do you even use money to trade? Or like is goods all free? You know, like. Yeah, that's probably a bigger question for people to think about.
- [61:02]
C
Okay, all right, so to wrap up, I just want to make sure that we hit on details for any potential applicants to the accelerator. You know, is there like a formalized application process and if so, how do they find that?
- [61:18]
B
We are launching a website called Eastworlds IO. I forgot, is it ezos.com or isos IO? Just be careful, don't get fish before we give you guys the more official link when it's launched. But this website is where there will be a application portal. Any teams who are interested, you can reach out to that. You can reach out to our virtual protocol Twitter or my personal Twitter as well. And we will have a chat. We have an entire ventures team that looks into great founders. We have an entire BD and ecosystem team that onboards founders into the ecosystem as well all the time. So we would love to welcome you. So if you are a builder, be it a digital agent or if you're keen to build on the humanoid side of things. Yeah, let's go and do some cool stuff together.
- [62:10]
C
Okay. And then as a reminder for being part of the accelerator, what, what are the things that they receive?
- [62:17]
B
Yeah. So as part of accelerator, there is going to be a free access to human. Right. Robots. There's going to be one dedicated per team. There is going to be access to research labs. So if you don't want to train your own VLAS or policies, we have people across different universities doing that for you. And the last bit is you get to work with us in person. So I think that's going to be quite fun.
- [62:44]
C
Okay. All right, well, just last question. So for you, you know, what do you view as success for the accelerator? Like what metrics will you be tracking?
- [62:55]
B
For us, it's at the very near stage, it's going to be a number of great projects. That come out of the accelerator. So think of it as like the
- [63:06]
C
next
- [63:08]
B
having a unicorn or having a couple of projects that will reach a $100 million market cap. That would be near term success. I would say longer term success is seeing the robots are being built in our incubator appearing on the news across the world, right? People be like, hey, look, we have this restaurant being fully automated by robots, and these guys are part of, you know, the East World Accelerator that I feel it's maybe something to smile upon.
- [63:43]
C
Okay, great. Well, thanks so much for sharing. Where can people learn more about you and virtuals and the accelerator and any other links that you want to share, just follow Twitter.
- [63:55]
B
We are most active there. Our handle is virtual protocol. Oh, sorry. Virtualsio. That's how this handle. And yeah, just follow that. It has like 260,000 followers there. So just. Yeah, that's where all the news happen. You'll never get lost.
- [64:14]
C
Okay, perfect. All right, well, thanks so much for coming on Unchained.