GateLive Roundtable Episode 11: AI + Web3 = Web4? — Exploring New Opportunities in the Crypto Sector from the Perspective of Agent Economy

“Gate Live Roundtable Discussion” is a Chinese cryptocurrency roundtable interview program created by Gate Live, broadcast every Wednesday at 8:00 PM. It focuses on the most discussed industry topics of the moment. The program invites core practitioners and frontline observers from fields such as blockchain, Web3, DeFi, the Ethereum ecosystem, stablecoins, and regulatory policies to engage in deep discussions in the live broadcast.

The roundtable features a casual, open, and authentic dialogue atmosphere, exploring market trends, industry disagreements, and key variables from multiple perspectives, helping the audience form clearer and more rational judgments amid complex market conditions and narratives.

This episode’s theme: AI + Web3 = Web4? — Exploring New Opportunities in the Crypto Space through the Lens of the Agent Economy

This episode’s guests: Well-known KOLs from the Chinese cryptocurrency community — Che Jiu, Hou Shan Ren, Grace

The content of this program is for informational exchange and opinion discussion only and does not constitute any investment advice.

(This content has been organized from the live broadcast replay, with text assisted by AI and appropriately shortened. For the complete content, please copy the link to view: https://www.gate.com/zh/live/video/731b83581579f056c8b780b19d384c42)

Host Jesse:

Hello everyone, welcome to tonight’s Gate Live Roundtable, I am your host Jesse!

A month ago, the founder of a cryptocurrency research project, Sigil Wen, published the “Web4 Declaration,” which sparked widespread discussion. Following that, the explosive success of OpenClaw allowed the market to witness the key leap of AI from “answering questions” to “autonomous trading and asset management.”

What truly shook the crypto community was a statement from Haseeb, managing partner at Dragonfly. He said — “Cryptocurrency may never have been designed for humans, but for AI.”

This statement sounds a bit counterintuitive, but upon reflection, it makes some sense: because the on-chain world is full of risks where “operational errors can lead to severe consequences,” a single mistake could result in permanent loss.

However, this highly stressful system for humans appears to be a clear, verifiable, and automatically auditable standardized process from the perspective of AI.

Thus, the concept of Web4 emerges, shifting cryptocurrency technology from “serving humans” to “serving AI,” making agents the new mainstay of the internet economy.

So today, we want to discuss a core question with our guests: When AI begins to have wallets and can trade autonomously, does that mean Web3 will come to an end, and Web4 will usher in a new beginning?

Today, we are very honored to have three industry veterans, Che Jiu, Hou Shan Ren, and Grace. Welcome to you all! Let’s first ask each of our guests to briefly introduce themselves.

Che Jiu:

Hello everyone, thank you for the invitation. I am Che Jiu. I entered the space at the end of 2023. I originally focused on GameFi and also built communities, but that sector isn’t doing well, so I started to transition. I now explore Alpha and study DeFi and yield farming, and I’m also operating on Twitter and very interested in AI. Thank you, everyone.

Hou Shan Ren:

Good evening, everyone. I am very honored to be invited to participate in today’s roundtable. I am an old hand in the space for 8 years, skilled in macroeconomic analysis, news interpretation, K-line analysis, and on-chain tracking, though I can’t claim to be an expert. I broadcast daily on Gate; please follow me if you can. Especially for those who enjoy spot contracts. I’ve also recently studied AI and am sharing from a learning perspective, not as investment advice. Thank you.

Grace:

Thank you, Jesse, for inviting me. I feel a connection with Gate; I entered the space in 2021, and Gate was the first exchange I registered with. My research interests are broad, including investment research, AI, prediction markets, etc. I am proficient in English and often host or participate as a guest in bilingual discussions.

I recently noticed that Gate has integrated Polymarket and GateClaw Lobster, remaining at the forefront in both the AI and trading domains. I hope we can grow together with Gate and look forward to the insights from today’s guests. Thank you.

Host Jesse:

Thank you once again to our guests for joining us. Let’s not waste any more time and dive directly into today’s topic discussion.

Earlier, we learned the basic concept of Web4. However, there is a fundamental disagreement: Is Web4 a natural extension of Web3 or a replacement? What do you think?

Che Jiu:

It is a continuation and upgrade, not a complete replacement.

• The best thing about Web3 is that it allows ordinary people to truly ‘own’ their assets without relying solely on large companies.
• Web4, on the other hand, extends that ‘ownership’ ability to AI — AI becomes our super assistant, helping us to manage money, invest automatically, and get things done, but the final decision-making power still lies with us humans.

1. Some believe Web4 is a continuation: AI is like the butler we hire, helping us manage assets and making it easier to use Web3.
2. Others say Web4 is a replacement: the cumbersome operations of Web3 (remembering private keys, fear of mistakes, high gas fees) are torturous for humans but perfectly suitable for AI, which doesn’t tire or mind complexity and can work 24/7.

In my personal view: The two are not in conflict; they are a progressive relationship. Web3 has laid a solid foundation for us (wallets, smart contracts), while Web4 allows AI to take the lead. In the future, there may be several times more AI users on-chain than real people, making the crypto world larger and faster. But humans remain the bosses; we are responsible for ideas and setting rules, while AI handles the hard work. This will create many more opportunities in the entire sector!

Hou Shan Ren:

I personally believe Web4 is an extension and surpassing of Web3, but it hasn’t reached the point of replacement; rather, it exists at a balanced point between extension and replacement. Why do I say that?

As everyone knows, the core of Web3 is simple: it allows us humans to own assets and use those assets and data to create value, invest, or manage finances, etc. What is Web4? Web4 essentially means the assets are still yours, but now AI Agents do the managing for you. They manage your wallets, automatically execute your trading plans, optimize your strategies, etc.

In other words, you no longer have to keep an eye on those tedious operations; you can just hand them over to AI. However, everyone needs to understand its essence: AI is fundamentally still serving us humans, meaning sovereignty remains with humans; the executor has simply changed to the AI of Web4.

So my view is that Web4 is not a replacement for Web3, but it is not merely an extension of Web3; it is an evolution after the extension of Web3, with the main participants shifting from humans to AI. But as I mentioned earlier, humans are still the ultimate subjects; it fundamentally serves humans, so humans are the ultimate owners and a so-called overseer. My understanding is like this. To use an analogy, for example, a car — we have the keys. Web4 is like the autonomous driving feature, but the car is ours, and the steering wheel is in our hands. It can drive, but we always control the steering wheel; if I say left, it goes left, if I say right, it goes right. That’s roughly how it is.

Grace:

I believe the answer is not an “either/or” situation, but rather that Web4 both continues the core principles of Web3 and, in its usage, represents a certain “replacement.”

First, let’s talk about continuation.

What is the most important principle of Web3?

In one sentence: Allow users to truly own their assets and data. It hands back “control” from platforms to individuals.

Web4 does not overturn this point; rather, it reinforces it.

However, Web4 introduces a key role: AI Agents.

In the future, you may no longer personally operate wallets, sign transactions, or manage assets; instead, your AI assistant will replace you in completing these complex operations.

So, from this perspective:

Web4 is a natural evolution of Web3 — upgrading from “users manage themselves” to “AI manages for users.”

The view of “replacement” also makes sense.

The complexities of wallets, mnemonic phrases, gas fees, signatures… but for AI — this is precisely the “perfect interface.”

AI is not afraid of complex processes, will not forget passwords, and certainly won’t click the wrong links.

This brings about a profound change:

The true “users” of Web3 in the future may not be humans, but AI.

Humans will only need to say one thing:

“Help me configure an asset portfolio” or “Help me participate in this DAO vote.”

The entity actually executing these on-chain operations is AI.

From this perspective:

Web4 indeed “replaces” the user layer of Web3.

It does not replace technology but replaces the role of “humans as direct users.”

The conclusion is:

Web4 is not merely a continuation, nor is it a complete replacement; it is a “role reconstruction.”

Web3 addresses the question: Who owns the assets?

Web4 addresses: Who will use these assets?

The future world will likely be:

Humans own assets, AI uses assets.

Whoever’s AI agent is smarter and more trustworthy will have a competitive advantage.

Host Jesse:

The previous discussion was very enlightening. If the logic of the continuation faction holds — AI is a proxy for humans — then a key question arises: What money does AI use? What are AI’s asset reserves? This leads us to our second core topic of discussion today: Will stablecoins and RWA become the currency and reserves in the AI era?

Hou Shan Ren:

I think it’s highly likely. Why? Because stablecoins are the core foundation of Web4. To put it simply, in the real world, when we deposit and withdraw money, we either use WeChat or have bank cards, etc.

First, you need to go through identity verification; that is, you need to provide an ID card, and it must be a unique one. However, AI does not require that; it does not have a unique identity verification, only this so-called stablecoin. Stablecoins also have a characteristic — because of the trading characteristics of AI Agents, they can handle more complicated transactions, such as high-frequency and small amounts, and they do so without emotion.

Traditional financial systems can’t handle this because you can’t pass KYC. If you take out too much money from the bank, you have to queue again tomorrow, right? This is impractical. So it’s fundamentally mismatched; therefore, in Web4, stablecoins are absolutely inevitable, and are even a very core infrastructure.

Then about RWA, can it become the so-called reserves of Web4? I think that’s also acceptable; I can agree.

RWA, as we know, refers to the tokenization of real-world assets. If a robot earns money and wants to store that asset, it would typically invest it or keep it in a bank, right?

But what can this robot do? It would have to buy financial products from exchanges, which still requires your consent. However, with RWA, it would be different. Once real assets are directly on-chain, it can analyze based on the real-world prices, and then according to its own funds, turn it into a sort of financial reserve, meaning you give it commands to let it earn money on its own.

For example, when we humans earn money, we want to buy cars, houses, or similar items. In the Web4 scenario, the AI agent is like a human in the virtual world, and it also buys cars and houses from its reserves, but it must correspond to on-chain assets. The combination of stablecoins and RWA as a so-called value anchor becomes a reserve, which constructs a very perfect currency system.

Grace:

Stablecoins + RWA are likely to become the “currency layer + reserve layer” of the AI era.

But this is not a simple replication of traditional finance; it serves an entirely new entity: the AI agent economy.

First, let’s discuss the first part: Stablecoins = “Machine Money” in the AI Era.

In the future, we will see many scenarios like these:

• AI automatically calls APIs and pays in real-time
• AI purchases data and trades computing power among themselves
• AI agents help users complete cross-border payments and subscription services

These transactions have a few characteristics:

High frequency, low amount, automated execution, globalization.

Traditional payment systems cannot support this kind of “machine-level settlement,”

while stablecoins precisely solve three core problems:

1. Price stability (facilitating pricing)
2. On-chain settlement (no need for trust)
3. Programmable payments (suitable for automation)

So I believe:

Stablecoins are essentially designed as “native currency” for AI.

The second part: RWA = “Reserve Assets” for the AI Economy.

If stablecoins are “cash,” what is needed behind them?

The answer is: Credit and reserves.

That is the significance of RWA.

RWA (tokenization of real-world assets) includes:

• Government bonds
• Real estate
• Commodities
• Income-generating assets

The characteristics of these assets are:

Stable, predictable, with real cash flow.

For AI, it does not just need to spend money; it also requires:

• Managing asset-liability sheets
• Configuring income portfolios
• Hedging risks

In other words, AI is not just a “consumer”; it will become an asset manager.

In this system:

Stablecoins are used for circulation

RWA is used for reserves and appreciation

Finally, let me make a more critical judgment:

Stablecoins + RWA are not just a combination of technologies, but a “new financial infrastructure.”

Whoever controls:

• The issuance and liquidity of stablecoins
• The on-chain capabilities and asset quality of RWA
• And the interfaces with AI systems

may become the next generation of “financial operating systems.”

In a nutshell:

Stablecoins solve “how money flows,”

RWA solves “what money is worth,”

and AI decides “how money is used.”

Only by combining all three can we achieve a true AI economic closed loop.

Che Jiu:

Yes.

First, regarding stablecoins: they are particularly suitable for AI use. Because AI needs to pay continuously, in small amounts, and automatically — for example, when two AIs help each other, they need to settle immediately.

Stablecoins like USDC are stable and fast, and they can operate 24/7 without sleeping. Humans have to think about emotions when transferring money, while AI can finalize transactions in milliseconds, making it ideal for high-frequency trading.

Now looking at RWA (which involves turning real-world properties, bonds, and government bonds into on-chain tokens): this will become AI’s “reserve.” AI cannot only hold volatile cryptocurrencies like Bitcoin; it needs stable returns to sustain itself. With RWA, AI can automatically invest its earnings, obtaining interest from real-world assets, and the entire process is fully automated and verifiable.

In simple terms, stablecoins are AI’s “daily pocket money,” and RWA is AI’s “pension fund.” Together, they can truly kickstart the AI economy, potentially larger than the current DeFi.

Host Jesse:

For our final topic, let’s discuss something more practical: In Web4, what risks should humans avoid, and what roles should they play to better participate in this technological tide?

Grace:

I think the biggest opportunity in Web4 lies not in the technology itself but in how humans collaborate with AI. Before participating, we need to pay attention to three types of risks:

First, we may relinquish control without a verification mechanism. In Web4, many operations will be completed by AI Agents, including trading, investing, signing contracts, and even identity management.

The problem is, if AI makes decisions for you, do we still have the final control? I think the example given by Hou Shan Ren earlier was very interesting and insightful, especially the analogy of AI as a self-driving car — we need to tell it how to avoid risks. Sometimes the agent may also be subject to attacks. So the first principle is: all AI agents must be verifiable, reversible, and auditable.

The second type of risk is data assets being invisibly centralized again. Web3 was originally meant to solve decentralization. However, Web4 may introduce a new problem, that is, computing power + models + data being concentrated in the hands of a few platforms. If our AI Agents are entirely dependent on a specific platform, then our control over assets is still being entrusted. Therefore, the second principle is: key assets and identities must be maintained on-chain or within transferable systems. Otherwise, we would simply shift from dependency on one platform to dependency on this AI platform.

The third type of risk is the diminishing of human capabilities. AI can help us with so many tasks: investment, decision-making, programming, and even socializing. Are we as humans only left to confirm decisions? In the long term, this could be dangerous, leading to a loss of judgment or true control. Therefore, the third principle is: humans must maintain decision-making authority and understanding, rather than completely outsourcing it.

After discussing the risks, what role should humans play in Web4? I think three roles are particularly crucial:

1. Become rule-makers. AI is merely an executor; one of the most important abilities for humans in the future is to design constraints on AI behavior.
2. Act as value judges. AI can optimize efficiency, but it cannot truly define what is most important. For instance, which assets are worth holding long-term, which projects deserve support — this value judgment belongs uniquely to us humans.
3. Become system owners. Future competition will not only be about individual capabilities but also about who owns stronger AI systems. In fact, everyone needs to build their own personal AI operating system.

So, Web3 allows humans to own assets, while Web4 allows AI to use assets. And humans must maintain ultimate control. That’s my sharing, thank you.

Che Jiu:

How can we better participate in this technological tide? I think the key is to first learn to avoid pitfalls.

What are the pitfalls:

1. The risk of AI going off track — it can be too autonomous; if it misunderstands your intentions, it may operate erratically.
2. The risk of asset loss — even if AI manages your money, you should give it minimal permissions and not grant too much at once.
3. Privacy leaks — AI requires a lot of data, which can easily expose your personal information.
4. Social issues — the high efficiency of AI could reduce some job opportunities.

How to avoid pitfalls?

1. Be a designer: We set goals and rules, telling AI which direction to go (AI excels at execution, while humans excel at thinking of the big picture).
2. Be a supervisor: We retain final decision-making power and check in regularly.
3. Be an innovative partner: We provide real-world demands and data while using AI tools to enhance our own capabilities.
4. Be an opportunity seeker: You can start using AI Agents like OpenClaw now to learn how to build your own small AI team; entrepreneurs can create tools and payment protocols for AI use; investors can position themselves in relevant projects early.

Hou Shan Ren:

First and foremost, we must avoid losing control. Just like the example I provided earlier, even if the car has autonomous driving, you cannot let it go out of control. You can allow it to drive autonomously, but only under the premise of ensuring safety, without the risk of losing control. A hypothetical loss of control occurs when you grant AI too much authority.

This is a bottom line, a core principle. You need to limit AI within a certain range of permissions to operate. It cannot touch the bottom line; anything that crosses the red line must receive your authorization to proceed. Therefore, we must first avoid losing control and establish principles for AI; what it can and cannot do. This is equivalent to the laws of the Web4 world.

Then there’s the risk mentioned by Grace regarding humans being replaced. We are the creators and supervisors, so we cannot be replaced. This risk exists; many people worry that AI will take their jobs or similar concerns. You can think about what kinds of jobs are not replaceable by AI — you should focus on more creative and valuable tasks. You can consider AI as your employee, and you need to plan behind the scenes to ensure it delivers results.

Host Jesse:

Time flies, and we are approaching the end of today’s discussion.

In summary: Regardless of whether you are part of the continuation faction or the replacement faction, one thing can be certain — cryptographic technology is shifting from “serving humans” to “serving AI,” and agents are becoming the new mainstay of the internet economy.

Is the “poor user experience” of Web3 a design flaw or a standardized interface prepared for AI? This question has no standard answer, but it is worth our continued observation.

Thank you to our three guests, and thank you to everyone watching. See you in the next live broadcast!