Retired electronics worker Fan Xinquan in Beijing has just begun raising a so-called lobster after he hopes that the AI agent he has been training will prove useful in organising his expertise in a particular field more effectively than chatbots such as DeepSeek.

The 60-year-old said at a recent event hosted by AI startup Zhipu to educate people on how to use and train the AI ‌agent, which has become viral in China, that it could actually help them “achieve many practical things with OpenClaw.”

Over the last month, OpenClaw, the ability to hook multiple hardware and software components together and learn from the data generated with far less human interaction than a chatbot, has captured the imagination of many in China, both retired people seeking side earnings and AI companies wishing to create new revenue streams.

Initially debuted in November, the tool has seen rapid growth and is currently ranked among the very most popular projects to have ever been hosted on GitHub, the most popular AI-powered developer platform in the world.

The most recent example of how a new technology has the potential to transform the world such that the second-largest economy has no checks and balances on its anarchic consumer use is the hype about the open-source, agent-controlling bot by Austrian developer Peter Steinberger.

“OpenClaw exemplifies a breaking point of open-source ⁠agents, much like DeepSeek is a breaking point of open-source large language models,” according to Wei Sun, chief AI analyst at Counterpoint Research.

This week, Nvidia CEO Jensen Huang claimed that OpenClaw is the next ChatGPT and that enthusiasm about the technology was on the rise, with Chinese tech stocks skyrocketing up to 22 per cent or more in recent weeks as firms unveiled an array of products built on the technology.

OPENCLAW ATTRACTS CHILDREN AND RETIREES

On Tuesday, at an event, Huang Rongsheng, who is the chief architect at Baidu smart device unit, Xiaodu, said that parent group conversations at his daughter’s primary school have been dominated by OpenClaw talks.

“They came to me, and my daughter told me: Dad, I see you raising a lobster daily,” he said. “Can I have one too?”

Another invitee to the Zhipu event, Bai Yiyun, expressed the desire to utilise the agent as a way of earning a side hustle when she retires.

“In fact, some people use it to purchase lottery tickets or to pick stocks, some use it to develop money-making apps or open e-commerce stores, but I do not know whether it brings them any actual profits,” she said.

Beyond the get-rich-quick schemes, a significant number of OpenClaw users expect their productivity to soar significantly, and subsidies of as much as 20 million yuan/2.8 million dollars a year are being offered to qualifying “one-person companies” by some local governments.

The OpenClaw frenzy is simply directly correlated with what the Chinese government desires regarding the AI Plus program, said Lian Jye Su, the chief analyst at tech research firm Omdia, referring to a national policy to integrate AI into the economy.

SECURITY RISKS AND TECHNICAL CHALLENGES

However, the initial burst of enthusiasm might yet dissipate, in particular, as token costs grow and regulators threaten security weaknesses. This week, Zhipu announced a token price increase on its new OpenClaw-optimised AI model by 20 per cent.

“Production is very poor: commoners are spending tens or hundreds of yuan, setting a stack of tokens on fire, and eventually, they may end up with a stack of useless data in their hands,” read a post on Rednote, a social media platform called Goodbye OpenClaw.

“This is not reaping the future,” it said, “it is being reaped by the future.”

The pervaporation of Chinese society and industry has taken its toll on Beijing, as an increasing number of Chinese institutions – such as government agencies, brokerages and universities – are now banning their employees from installing OpenClaw after being warned against it by their regulators.

A state-owned newspaper article released last week by the state-run People’s Daily, the mouthpiece of the ruling Communist Party in China, called on the government to establish and preserve the safety bottom line to see that innovation does not go off track or come off the rails with OpenClaw.

“People in Beijing clearly view AI as a strategic priority and would like Chinese companies to commercialise within a short time frame,” said Rui Ma, the founder of the Tech Buzz China newsletter.

“But it also desires to be deployed in such a way that it remains legible, secure and politically controllable… the worry is the completely uncontrollable and chaotic diffusion that will inflict harm.”

Data security professional Li Hongxue, with a finance company, found the contrast between local government actions and warnings by the central government “contradictory.”

“It cannot be stopped, but its development,” she added, “must keep pace, so in that regard, this can also be an opportunity to (my) field.”

The second question is whether the agent will have an easy time moving between applications and devices that are under the control of a large mix of companies, which are occasionally in rivalry with each other.

During the Baidu event on Tuesday, an employee ordered coffee on a McDonald’s application by voice command, via a Xiaodu smart device, which is facilitated by an OpenClaw agent.

It was nearly two minutes before it was time to pay the order.

“As you can see, I did actually just issue a bare command, but to get the entire delivery through there is actually a lot of work being done behind the scenes by Xiaodu and the lobster,” said the Baidu employee.