Return of the metaverse. Exactly. What does that look like? Okay, so let me first start off. How many people here still remember CryptoKitties? Raise your hand. There we go. All right, fine. So the reason why I'm putting up that slide is this was in 2017, and that was a time when blockchain, when Ethereum basically melted, when there were only a few million dollars worth of digital cats being sold. Of course, today with stablecoins and with advancements on blockchain and all the main institutions like, you know, Franklin Templeton and BlackRock, you know, you've got basically trillions of dollars of transaction volumes. Basically, blockchain has become scalable, and we have also now something like almost 800 million people in the world who own crypto. But we do have a little bit of a problem in the space: only less than 10% of them are actually onchain. So DeFi and the whole onchain economy has been actually not growing in comparison to people owning crypto.
But perhaps one way to think about this is not that crypto isn't relevant. It's that perhaps the human experience of crypto remains very, very difficult. I mean, for those of you in crypto, how many of you are still using MetaMask? It's a great product, by the way, but it's harsh. My mom will never use it. My daughter won't use it. They can't handle the seed phrases. It's kind of tough. And even 10 years in, we still haven't matured out of that kind of scenario. But what if all the technology that we've seen built on blockchain was in fact not meant for humans, and that we as humans were basically the PC for agents to come? Let me sort of elaborate a bit first. We think that blockchain is the native settlement layer for the AI economy. You've probably heard this in a number of talks. I won't sort of talk about it too much, but it's natural for them. Basically, the blockchain was not designed for human speed. It was designed for machine speed. It has fault tolerance. It's decentralized.
Trillions of dollars of commerce can actually happen through, you know, X402 payments or direct transactions between agents through essentially a direct settlement layer onchain that is direct, and it can also do it cross-chain. For an agent, it's super easy. The second thing is that tokens actually are naturally kind of digital bearer assets for the machine economy. So right now, we initially started with things like memecoins and had some fun with virtually digitally native tokens and built our own network effects. But today, tokens are representative of stocks and equities and commodities and other kinds of forms of yield. They are still bearer assets, except they are digital versions of that. But are the principal buyers going to be humans or, in fact, agents? And we think that's actually the way in which agents are going to transact through this form of digital value and currency, and agents are naturally users of these types of tokens.
Traditional finance cannot bank an AI. It's going to be a long time before a bank has the ability, if ever, to give an agent a bank account. But we know that agents are going to get smarter. They're actually getting more intelligent. Their IQ is getting better at some tasks, eventually at most of the tasks that we do as people. So what's missing? They need to have a form of power, which is money. How do they get it? That's not going to be a bank account. It's going to be the same way that people who didn't have a bank account or couldn't get one in places like the Philippines now, through a crypto wallet, basically become bankable in the entire world of DeFi. All the vaults, all the yield, and frankly all the DeFi banking capabilities become available to them essentially for that machine economy. So in simple, fast conclusion, you could say that the blockchain and crypto is the perfect financial system for machines designed for 24/7 operations.
Just think, for instance, why are people so keen on trading tokenized assets over the weekend? Because a regular stock market is closed, but the blockchain continues to operate 24/7. There is no circuit breakers. There's no 9 to 5. There's no holidays, no Christmas, just go. What are the systems that can trade and transact and basically move money and value 24/7? Not a human. We still have to sleep. It's going to be an agent. And therefore, this is the system designed for that. Now, what that means, for instance, though, is the whole world becomes more agentic. We're no longer going to be discovering the web in the traditional way. It won't be HTTP, it's going away. We're all going to be searching it through APIs, CLIs, MCPs, and of course tokenized types of assets. I mean, think about the last time when you were on Gemini or ChatGPT and they gave you a bunch of websites. Did you actually click through to the website, or did you just take the summary of what it said and say that was sufficient?
So by many accounts, in fact, advertising revenue for a lot of these websites that relied on Google clickthroughs has dropped by 80%. So you have to be agent-ready and agent-readable. But in the backdrop, what happens is that agents are going to be doing all these direct transactions with each other. And they're not going to be doing just specific X402 payments. They're going to sort of do other forms of transactions, new systems, whichever. They can help plan your trip, and they're going to use different agents. It could be microtransactions, you know, things that you can't do with a credit card, but things you can do onchain, like the fraction of a cent of a, you know, microfraction or something. They can transact and basically sort of do a transaction and basically purchase something, and this will happen more and more naturally. And of course, the way to audit that is going to be the blockchain. So in this environment, you basically no longer have the traditional advertising attention economy.
Everything becomes frictionless. Everything is running in these microtransactions. Everything is heavily customizable because the agent can code. Now, let me elaborate on that a little bit. When an agent can code everything quickly and easily, you no longer need to have apps and software as you see it. You're going to have an agent deliver the application just how you want it in your inbox or in your WhatsApp or in your Telegram, right? And they're going to customize these applications in the way that you already sort of want to experience it. And they can speak and be compatible and portable in other ways. In fact, the prediction is that 80% of apps and software as you know it is going to go away.
All right. Now, let me quickly move into sort of the whole metaverse thing, since that was a big topic here. One of the things that we thought back in 21-22 is that we're going to spend a lot more time online with these goggles and other things because it was COVID. We all believed that, you know, there's a deadly virus out there. We will no longer be able to meet people as freely as we wanted to. So we're going to go all virtual. Turned out that that was not quite true. However, does it mean that the metaverse is dead, as a lot of people talk about? Actually, we think quite the opposite with agents. Maybe it wasn't about us spending more time online, but instead it was about AI agents coming to life in the virtual plane, which are these AI agents, but that can do things for us in real life. In other words, it's still the metaverse, but instead of us being in it, they are coming to us.
And think for those of you who are using agents today, they're handling your calendar. They're doing flight bookings. They're maybe saving you money somewhere, making money for you. Even the crypto concepts of things like play-to-earn actually have come to life. Except the play-to-earn construct isn't farming some token. It's actually creating real value in real life, effectively doing things like, I don't know, building websites or saving you money, maybe help you walk the dog, all that kind of stuff. AI agents are reshaping the physical world, but they are actually metaversal in construction. So in that sense, and that's kind of what we're pushing with Hello Minds, which is our own agent platform that has blockchain on the back, AI agents become a cast of characters. Everyone is generally talking about AI agents in the way where it's a single agent that manages everything for you. But actually, do you really want one agent managing everything for you?
If they manage everything for you, can you share that agent with your partner or your work colleagues? You probably wouldn't do that because it knows everything about you. But you might want to compartmentalize that: one agent for your family, one agent for your friends, one agent for work, right? Maybe for specific tasks, right? That's what we think is going to happen. It's much easier not just for the context management for the agent, but also for context management for us as humans. So agents become a cast of characters, much like a video game, whether it's like The Sims or some kind of civilization game, except they're ones that have true intelligence that can help you in real life. So some examples that you can do with Minds and with agents: I've got one that lowers my grocery bill on daily purchases in Hong Kong. That's kind of fun. You know, a price of eggs, it seems cheap, but when you do that every day, it compounds to a value of thousands of dollars.
You have agents that can handle your investments. You have agents, and this is one of my sort of personal favorites, that can handle your Instagram account. If you're a small business, it's very easy for them to post videos and make postings for you and so forth. And so the way to think of the entire environment for AI agents is basically like a playable home for your personal AI, instead of thinking of it as a tool. It's about social, it's about entertainment, it's about fun. It's like playing a game, effectively like the game of life, if you will, except the outcome is real impact for real tasks, real life progress. What you can do essentially with Hello Minds, and frankly other agents as well, except this one is one that is multi-agent, and the environment, if you download the app as well, is one where you can see it in a gamified kind of way, which we think is a way that most users will translate easily to.
All of you are here probably entrepreneurs, you're founders, you understand the space really, really well, so you already know what you want. But most people, what we found is when they get an agent and they ask them the question, something like, 'What would you like me to do?', they don't know what to ask because they don't know the question to ask because they don't think of hiring someone, right? So this entire experience that we see in the gaming world or on virtual worlds, the simulation that we've experienced over the last few decades, which are really just simulations of the virtual world of essentially a gamified experience, was doing what it was in many ways mimicking real life. It was mimicking social real life as well as activities in real life. You learn about economics a little bit. You learn about trade. You learn about friendships. You're learning about game actions. Now that is coming to the physical world in the form of these AI agents and everything that we've been trained for.
And before we start saying this sounds crazy, you know, this is a game, this is very different from real life, I'll remind you that Nvidia, which of course started from gaming, was originally doing simulations of these 3D environments and basically real-world simulations in a gamified way before it became the engine for this entire AI and, frankly, Bitcoin mining community at start.
So we believe there's going to be hundreds of billions of agents, at the smallest estimates of 50 to 100. And the reason we believe that is if you ask yourself how many agents you think you're going to have in the next 3 to 5 years, the common answer is 3 to 5. If you're in San Francisco, it's a few hundred. Depending who you talk to, to an engineer, maybe it's a few thousand. I personally now have 400 agents. I don't think you need 400, but you'll definitely need more than one. And so by that account, we have 6 to 7 billion people online. There'll probably be 7 to 8 in the future. Multiply that by roughly 10 agents average per human, you're going to have somewhere between 50 to 100 billion agents minimum floating around the world. But each of them are actually going to have a wallet. They're going to be smart. They're going to trade and transact.
They're going to essentially be moving us towards an environment where agents are more populous in this new web than humans. And that means some of the concepts that we've said in crypto, even things like DAOs, decentralized autonomous organizations, are manageable with agents in a way that was difficult for humans because of our attention, because of the complexity. But actually a DAO could be very easily managed by an agent, and that you could build a new form of agentic consensus, if you will. And so what we believe right now, we're kind of at this early stage where AI is a tool. It's what you use mostly as a chatbot. You ask it a question, it gives an output. We're moving, it's a small percentage of the world that is moving AI as an autonomous agent, you know, with a little bit with Claude Code, OpenClaw, and now with Instinct and Muse, and of course Hello Minds. But it's very early stages. This AI as autonomous agent is really literally only in the collective millions of users.
It's very small in comparison where we are using AI as a tool. But where we're getting to very quickly, when they're on the blockchain, is they become economic actors. More onchain activity and trading, and, you know, basically transactions will happen through AI agents. And eventually, when they become economic actors contributing real new GDP, they become asset holders. And the crazy thing as it may sound is that, you know, how many people who work for Uber, who are drivers for Uber, actually care that, you know, they're getting paid by a machine or by a human? They never even get to interface with the person who's making the payment. So if an AI is paying you or an agent is paying you as a human, do you actually care? Does it even matter? Protocols are paying people today already through, you know, token airdrops. Does a user care? They don't care, right? So we will expect AI as economic actors and asset holders to also be the ones who will be generating forms of employment for humans as well, and probably in the digital asset space, potentially amongst the biggest asset holders, whether they're working for someone or perhaps even working for themselves, who knows.
Now I want to close, I have a couple of minutes, I want to close with an important topic about agents. And, you know, the future of what we would describe the new metaverse is around digital sovereign agency. When an AI agent can do so much for you, we think decentralization and digital assets become ever more critical. And the notion of owning and having digital property rights becomes critical because the question is no longer who owns your data. It's who owns the agency of the agent. So news came out. These are the big platforms in the world. We live in a kind of digital feudal environment. One thing to think about is for those of you who have heard of Muse and are thinking about using an AI agent, how many of you would trust Facebook with all the data that this AI agent has? Raise your hand.
Okay, none of you. And that's the right answer. You wouldn't trust them. The same way that they remove you, we looked at the terms of service. They have the right to remove your agent at any time. They can use your data to train, and they do not guarantee that when you delete the agent, that the data that you have cannot continue to be used for training. That's in their terms of service. You can take a look at that. And that is true for all of these platforms. So it's very different when you lose your social media account for a couple thousand or tens of thousands of followers. It's painful, maybe as a business, very painful. But it's something different if you have an agent that basically works for you, does things for you, you depend your life on, has your bank account details, knows everything about your family, you use them for years, and suddenly it goes away. Not to mention the data security and privacy that happens.
Decentralization therefore matters incredibly, not only because agents don't mind and transact with each other, because that's actually the mechanism in which we can actually own our agents. In the same way that you could own Bitcoin in a decentralized manner and have sovereign ownership, you can now own an agent in the same particular way. And I'm out of time, so I'm going to basically give you a quick, quick final thought. These are the safety and privacy infractions that Facebook has basically had, or Meta, which is in the billions of dollars, right? And so here again, if you want to have a future in which we can truly own this next web, we need to own our agents. It cannot be on Facebook or news. You got to do it on something that is on the blockchain that you can basically have true digital ownership on. Thank you very much.
