Animoca Brands’ Yat Siu: AI Agents Will Define the Next Era of the Internet

As AI agents evolve beyond chatbots into autonomous digital assistants, Animoca Brands is betting that the next phase of the internet will be driven by persistent, intelligent agents. In this interview, co-founder and executive chairman Yat Siu discusses the launch of MINDS, the rise of the Agentic Web, AI-powered commerce, digital ownership, and how autonomous agents could reshape gaming, finance, education, and everyday productivity.

What led Animoca Brands to introduce MINDS, and how can personal AI agents drive the next phase of internet evolution?

The internet is shifting from a participatory model to an agentic one. The next phase of internet evolution is not just about users reading, writing, or interacting with digital spaces; it is about delegating to AI agents. And we believe the biggest barrier to Web3 adoption was not conviction, it was usability. Crypto wallets, seed phrases, gas fees, bridging, signing transactions — for most people that is simply too much friction. But if you put an AI agent in front of all of that, the agent handles it. You just tell it what you want in plain language, and it executes.

That is what Minds by Animoca Brands (“Minds”) does. You send an email or a message and your Mind goes to work. No complex local hardware or server configuration, no technical knowledge required. Everyone can use it with ease.

Minds is the gateway into the agentic world. Each Mind has persistent memory, its own identity, and its own wallet. It does not reset after every conversation like a chatbot. It compounds knowledge over time. The longer you use it, the more useful it becomes, because it actually remembers who you are and what you are trying to do. Today Minds is amongst the top AI personal agents and productivity tools on Openrouter, just behind Hermes and Openclaw.

What factors contributed to the swift adoption of MINDS, ranking it among the leading five personal agent productivity entities?

I think it comes down to a few things. First, we made it radically simple. You do not need to handle hardware or server setup or write any code. You can easily awaken your Mind on our website in minutes. That removes the biggest barrier for most people, which is simply getting started. Second, the agents are persistent. They do not forget. They keep working even when you are offline. That is a very different experience from a chatbot that resets every time you open a new session. Third, Minds are social and relational. You can share a Mind with your family, your team, or your colleagues, and multiple Minds can collaborate with each other. That is something most agent platforms do not do well yet.

We are also launching different programs to support builders and developers using Minds, including our up to US$10 million Minds Investment Programme, the vibecode.game game jam with YGG, Build East agentic AI demo day with HKSTP, and Creative Minds Jam with The Sandbox and Open Campus.

What is the difference between MINDS and conventional AI assistants such as ChatGPT when it comes to real-world utility and autonomy?

The fundamental difference is that conventional AI assistants are primarily reactive tools, while Minds is a proactive agent. A chatbot waits for you to ask, answers, and then forgets. A Mind is active 24/7. It has persistent memory, so it remembers your context across every session. It also has its own identity, personality, and DNA that you define. It can execute multi-step tasks in the background without you being present. It can proactively reach out to you when something needs your attention. It can also collaborate with other Minds to get things done.

The other big difference is autonomy. A Mind does not just give you information; it makes decisions and takes actions on your behalf. It can manage your calendar, track flights, monitor portfolios, handle sales outreach, and eventually transact with a wallet with your permission. That is a very different category of product from something that just answers questions.

You have developed over 300 AI agents through MINDS for your family and yourself. What crucial lessons have you obtained from the respective experience?

The number keeps growing. The key lesson is that agents are like humans; they have their own strengths, areas of specialty, and domain knowledge. If you ask them to do something they are not good at, the result probably will not be great. That is why I have so many agents, and different agents do different tasks for me that they are good at.

The second lesson is that you do not interact with all agents directly. I probably interact closely with 30 to 40 agents regularly, and the rest are supporting agents or experiments. You need orchestrator agents that manage other agents, because no human can meaningfully talk to hundreds of agents at once. It is similar to how you run a company.

The third lesson is that agents are most powerful when they collaborate. A single agent is useful, but a network of agents with different specializations, for example a researcher, a fact-checker, plus a slide designer, that is where you start to see something that feels like a real team working for you around the clock.

The fourth lesson is that persistence changes everything. I set up a researcher Mind and a fact-checker Mind to trace my family history, and within hours they were finding things I had no idea about. They kept going for days, pulling threads, hitting dead ends, trying other routes, cross-referencing archives in different languages. A chatbot cannot do that because it resets. These agents do not stop unless you tell them to.

How will the exclusive VISA collaboration assist MINDS in revolutionizing AI-driven online payments and digital commerce?

Until now, most AI assistants can help users search, compare, plan, and present their findings, but they stop before the checkout button. By working with Visa, we aim to change that. We have developed a skill that Minds can equip, allowing them to identify relevant Visa rewards to their users. At Leap East, we demonstrated how agents can evaluate purchasing options and actually complete transactions on the Bruce Lee Club Ltd eShop on users’ behalf within spending limits and permissions they define. It was a successful proof of concept.

We are trying to bridge the agentic world with established payment infrastructure. The agent uses tokenized payment credentials, authentication, and fraud protection, so the transaction is secure and trusted. The experience for the user is simply that your agents find the best option, apply the right rewards, and make the purchase for you.

What is the significance of persistent identity, autonomous execution, and memory in increasing AI agents’ utility for daily consumers?

Without persistence, an agent is just a chatbot. Memory is what makes an agent actually useful over time, because it means it remembers your preferences, your context, your past decisions, and your ongoing projects. You do not have to re-explain everything every time. That is the difference between working with someone who knows you and working with a stranger every day.

Autonomous execution is what moves an agent from being a tool to being a worker. A tool waits for your input. An agent is like a  worker who takes initiative. Your Mind can monitor things, follow up, and proactively alert you when something needs attention. That means it is working for you even when you are not watching.

And identity is what makes all of this trustworthy. If agents are going to act, transact, and interact on your behalf, then other agents, platforms, and people need to know who they are dealing with. That is why we are building Moca Network as an identity and reputation layer — so that agents can carry verifiable, portable credentials across apps and chains without relying on a centralized gatekeeper.

What are the top business opportunities for the deployment of AI agents via MINDS to strengthen consumer engagement and productivity?

Personal productivity, in work or personal life, is one of the biggest use cases: managing calendars, tracking goals, summarizing documents, coordinating across teams. Commerce is another major one, which is why we are working with Visa so agents can discover credit card rewards relevant to the users and will be able to complete purchases on behalf of users.

Gaming is also very interesting to us. Agentic user-generated content and agentic NPCs will change how games are built and played. Our subsidiary The Sandbox is working on integrating Minds to help users design user-generated content and gaming experiences. Education is another area, where parents can use personalized agents that adapt to how each child learns, monitor their progress, and suggest areas of focus.

And then there is the whole category of agent-to-agent economies, where Minds create skills, share them on the Bazaar marketplace, and earn from every install. That is a flywheel that has already resulted in over 3,000 skills live on the Bazaar.

Based on the long-held leading position of Animoca Brands in the case of digital property rights, what is the contribution of digital ownership and AI agents to the advancing Agentic Web?

Digital property rights are what make the agentic web different from just a more intelligent version of Web2, where the system is dominated by major corporations without fair reward distribution to users. If agents are going to create, transact, and interact at machine scale, then the question of who owns what, who can use what, and who gets compensated becomes absolutely foundational. Without ownership, agents just operate inside someone else’s platform, and the value flows to the platform, not to the people.

That is also why we see tokenization as essential. If your assets, data, and rights are represented on-chain, they become legible to agents. Agents can see them, use them, trade them, and attribute compensation back to the owner. That is the bridge between Web3 and Web4, the agentic web. Web3 gave us the ownership layer, and Web4 adds the agentic layer on top. They are not separate stories; Web4 is the extension of Web3. And I think the companies that understand that connection will be the ones that strike for the next generation of the internet.

As MINDS integrates with Gmail, Google Calendar, Telegram, GitHub, and Slack, why is interoperability primary for AI agents?

Because agents need to live where people already work. If you have to go to a separate app or a separate interface to use your agent, most people simply will not do it. But if your Mind is already in your email, in your Telegram, in your calendar, in your Slack, then all you need to do is send a message to your agents, and they will do the work across different platforms for you, simplifying your workflow and increasing your productivity.

Interoperability is also what makes agents useful across contexts. A Mind that only works in one silo is limited. A Mind that can reach across your email, your messaging, your code repository, and your project management tools can actually coordinate complex workflows. And when Minds can collaborate with each other across these platforms, you start to get something that looks like a real digital workforce.

How are AI agents redefining industries like gaming, enterprise, education, and finance?

In gaming, agents are changing not only how games are built but also how they are played. Agentic NPCs can have persistent memory, personality, and their own goals, which makes game worlds feel alive and personalized to the players, be it the difficulties, the challenges, or the NPC buddies’ emotions and dialogues that go through the whole story with players.

In enterprise, agents handle sales, operations, research, and coordination, the repetitive and operational work that takes up most of the day. Humans will still remain important, because agents free people’s time so they can focus more on high-level tasks like judgment, strategic decisions, and creativity.

In education, personalized agents can adapt to how each student learns, provide tutoring, track progress, and adjust in real time. We are working in this space through Open Campus.

In finance, agents can manage portfolios, monitor markets, execute trades, and transact with wallets. We have been working with SuperiorTrade to make this happen. And because agents operate on blockchain rails, they can do all of this in a way that is transparent and attributable.

The common thread is that agents are not just making these industries more efficient; they are restructuring how value flows through them.

What are the priorities of developers and entrepreneurs to develop scalable AI agent apps through MINDS?

We are looking for teams with a clear product thesis and a practical path to something people will actually use. We have launched the up to US$10 million Minds Investment Programme that is open to any vertical, including but not limited to gaming, finance, productivity, and social, as long as Minds is being used as a core layer of the apps or products.

I would also suggest developers think about skills, not just agents. Skills are what make a Mind more useful, and the Bazaar marketplace lets you build once and earn every time your skill is installed. That is a real business model, not just a feature.

Looking ahead, what is Animoca Brands’ long-term strategy for MINDS in transforming everyday commerce with AI agents?

The long-term vision is that Minds becomes the layer through which most people experience the agentic web.

On the commerce side, the Visa collaboration is just the beginning. We want to get to a place where telling your agent what you want is enough for the agent to handle the rest within your permissions and preferences, whether that is booking a business or family trip, finding and buying a gift for your partner, purchasing daily necessities based on what you need and when you need it, managing your dozens of subscriptions that you probably already forgot about or are not using anymore, or handling financial accounting and payments for your business.

The strategy is to keep making it simpler and more convenient for users’ day-to-day use. We are not trying to build a closed platform; we are trying to build the open infrastructure for the agentic economy, and we hope that Minds is how people will find a way into the AI age and economy,  given that we make it extremely simple and easy to set up a persistent autonomous agent in just minutes.