The OpenClaw Frenzy

At the start of 2026, an AI agent called OpenClaw swept across the world and became the third breakout AI product after ChatGPT and DeepSeek.

A deep dive into Clawbot: why did it become the first breakout product of 2026?

By March, OpenClaw had overtaken Linux as the most-starred project in GitHub history. This AI agent, represented by a little lobster, made history in only three months.

The Shrimp Craze Goes Mainstream

Unlike assistants such as Doubao and Manus, OpenClaw is a completely open-source and immature project. Its founder has even admitted that it was built with heavy AI participation and started as something he made for fun. Look closely at those two facts and an astonishing conclusion appears: OpenClaw has no real technical moat and offers no substantive breakthrough beyond the language model itself. So why did it spread across the world so quickly?

I think it was the inevitable next step for AI at this stage—a result of timing, circumstances, and a perfect fit with the public mood.

Why It Broke Out

First, we need to understand what OpenClaw can do. It shifts the goal from answering questions to “doing things for you.” Its reach is broader than Doubao’s, and it feels more like a trustworthy assistant than Manus. OpenClaw escapes the browser and the chat box: it can take over social accounts, set calendar reminders, and operate a local computer through automation. For ordinary people, that image is powerful because it turns the abstract idea of an AI assistant into something concrete: I could have a digital employee.

This raises a contradiction. Why did companies with leading models, such as OpenAI and Google, not build this first, while an ordinary programmer did? The idea of letting AI control a local terminal has actually existed for a long time—in phones such as Doubao’s, and in Gemini built into Android. Their biggest obstacle was not technology itself but the unavoidable question of the AI era: information privacy.

The premise of OpenClaw as a “personal assistant” is access to local information and permission to call tools. Any company launching such a product would face serious public scrutiny. When data sits in a company’s hands, people naturally worry that their user profiles will be collected. A mature company cannot easily pull itself into a whirlpool of rights violations; that is a commercial calculation. As an entirely personal open-source project, OpenClaw did not carry the same concern. Its users chose to use it freely, its APIs came from third parties, and the skills and MCP servers installed for the model were deployed manually by each user.

So OpenClaw stood on the geographic advantage of a strong disclaimer, the timing of language models surpassing people in many domains, and the human advantage of matching the public’s fantasy of AI. Its sudden popularity followed naturally.

The Myth-Making Fails

But once OpenClaw escaped the programmer bubble and the “shrimp craze” reached the public, its supposed strengths—being immature and open source—became fatal weaknesses. People who did not even know what GitHub was felt they had to learn AI vocabulary, tinker with local configuration, and race to “keep up with the times,” as if they had found a lifeline in a fear-of-missing-out wave.

Large internet companies, including Tencent and ByteDance, along with opportunists who saw the opening, quickly promoted OpenClaw’s ability to connect to everyday social platforms and its supposed ease of use. Tutorials, installation services, and token reselling turned into a new side business—the classic moment when the people selling the shovels get rich first. OpenClaw became a giant AI bubble, mythologized as a futuristic assistant that could do anything. That image encouraged emotional spending and made the AI farce grow louder.

As people understood the product better, its weaknesses became impossible to hide: burning through tokens, privacy leaks, weak model capabilities, chaotic source code, and skills whose safety was impossible to verify. The lobster was quickly pushed off its pedestal. After uninstalling it, everyone rushed to criticize it as a way of expressing anger and disdain. It was a textbook case of failed deification: attention eventually turns against whatever it helped build, especially when the object was never perfect.

A Rational Way to Use OpenClaw

As a product with an unprecedented number of GitHub stars, OpenClaw clearly has unusual strengths. AI agents will be one of the main ways large-model technology reaches the real world over the next few years. The rise of these assistants will also help small companies take shape quickly. They can erase information gaps at high speed, just as social media once shook the world of print, and accelerate everyday workflows dramatically.

Rejecting technology outright is not useful. But getting trapped in the whirlpool of an immature product is just as wasteful. For an ordinary person, the only sensible answer is to wait for a product that is genuinely mature, trusted, and widely accepted.

Update, July 6, 2026: The once-inescapable OpenClaw finally disappeared from the conversation in June, replaced almost perfectly by Claude Code and Codex.