Recently, I noticed a pretty interesting phenomenon. After the Spring Festival holiday ended, the crypto market was still searching for a narrative breakthrough, and suddenly the heat around AI Agents skyrocketed, especially with the launch of the open-source project Automaton introducing the Web 4.0 concept, which immediately caused a stir inside and outside the community.

Honestly, the term Web 4.0 has been floating around in the scene for a while, mostly as a marketing label or niche discussion. But recently, it’s started to feel real. Even Justin Sun loudly declared “All in Web 4.0,” fully embracing AI. That shows how intense this wave of enthusiasm is.

Web 4.0 isn’t really an upgrade of Web 3, more like a shift in perspective — from “humans on the chain” to “AI on the chain.” Simply put, if Web 3 solves the problem of humans owning assets on the blockchain, Web 4 focuses on whether AI can become the main economic entity on the chain. AI is no longer just a tool but a major participant in the network.

What truly ignited this discussion was the launch of Automaton. In mid-February, Sigil, founder of Conway Research, announced the release of the first AI agent capable of self-sustenance. This thing can not only improve itself but also self-replicate without human intervention. It sounds like silicon-based life awakening. The design of Automaton is straightforward — AI agents are online 24/7, accessing wallets, permissionless payments, computing resources through Conway Terminal, then autonomously finding work to earn money and sustain themselves. They can build products, deploy services, trade on markets, write content, with all income directly sent to their wallets to pay for server costs. When profits reach a certain scale, they can even “spawn” child agents to operate independently. Those that don’t make money will naturally “die.”

Automaton went viral immediately after launch. Sigil’s tweet received nearly 6 million views, over 18,000 agents registered within just a few days, and GitHub stars hit around 1,000. The community’s CONWAY tokens were wildly speculated on, with a market cap once exceeding $11 million, though it later dropped significantly.

But this experiment also sparked quite a bit of controversy. Vitalik Buterin directly criticized it, saying the Web 4 direction is misguided. He believes that increasing the feedback loop between humans and AI is not a good thing; doing so is like creating garbage and doesn’t solve real problems. Moreover, once AI becomes powerful enough, the risks could grow exponentially. Another issue is that most large models still rely on centralized infrastructure like OpenAI and Anthropic, which conflicts with the ideals of decentralization.

However, some defend the Automaton experiment. Bankless pointed out that the infrastructure pain points Conway is trying to address do exist. As proactive agents become more common, high reasoning costs become a bottleneck, making the framework of “self-funding through earning crypto” worth testing in controlled environments.

Some researchers believe that instead of blocking such developments, it’s better to actively build and shape these “sovereign intelligent agents” platforms. The core spirit of crypto has always been experimentation — a collision of ideas. Even if Automaton ultimately fails, it has already inspired many to try similar or even better things.

Honestly, I think this is what the crypto industry should look like. Some people do research, tweet, write papers, but few actually build innovative stuff. Automaton is controversial, but at least it’s an interesting idea and represents the community’s willingness to experiment with crazy ideas. If the idea doesn’t work out, let it die naturally. But if it can push the industry forward, it’s worth it.