OpenClaw Craze Fuels Beijing AI Hustle

Beijing coder turns OpenClaw into a 100-employee AI services machine.

A single open-source tool is spawning a new class of AI services, right where you’d least expect it: a street-front hustle in China’s tech heartland. OpenClaw, an open-source AI assistant that can take over a device and autonomously complete tasks for a user, has spawned a fast-growing service economy. By January, Feng Qingyang, 27, started offering “OpenClaw installation support” on Xianyu, a popular secondhand marketplace, remote and instruction-light. By February, he had quit his day job. Today his side operation is a full-blown enterprise with more than 100 employees and a track record of roughly 7,000 orders, each priced around 248 RMB (about $34).

The scale is striking, but the pattern is even more telling. OpenClaw’s appeal is not just the tool itself; it’s the appetite for a low-friction, high-velocity AI services layer that can be deployed, installed, and managed with minimal coding. In a market where “DIY AI” is moving toward turnkey agents that can autonomously perform tasks, the revenue model appears deceptively simple: a high-volume, low-touch service economy built on enabling others to deploy autonomous agents quickly, remotely, and with limited technical overhead. Feng’s story—from solo tinkerer to manager of a distributed workforce—reads like a stress test of what open-source AI democratization looks like when it meets real-world demand.

For practitioners, the OpenClaw episode is a microcosm of several hard realities. First, scale arrives through labor-light, service-heavy models. The fact that 7,000 orders have come in at roughly $34 each underlines a business model that survives on volume and efficient, remote onboarding. Second, risk and governance rise with every new installation. An AI tool that can take over devices—however well-intentioned—creates a rich seam for misuse, liability, and security questions. Third, quality control becomes a management headache at scale. A 100-person team must implement standardized processes for onboarding, safety checks, and incident response, or risk inconsistent outcomes and user distrust.

From a product perspective, this quarter’s lesson is that open, flexible AI “agents” are not just a research curiosity; they’re a shifting service layer that can outpace traditional feature ships. Industry watchers will be watching how developers and platforms respond: will there be stricter licensing, guardrails, and usage boundaries around autonomous device control? Will regulators push for documentation, auditing, and safer onboarding flows? And how quickly will the economics of impulsive, on-demand AI services push toward standardized, repeatable playbooks that can be scaled without sacrificing safety?

Two to four actionable takeaways stand out for teams racing toward real-world deployment. First, guardrails and auditing matter as much as capability. Agents that can autonomously act on devices demand robust safety checks, user consent workflows, and auditable logs. Second, scalable service models require disciplined operations: training pipelines, quality assurance, and clear SLAs to keep up with demand without burning out workers. Third, the economics favor platforms that can combine low marginal costs with high-volume demand, but only if they can sustain reliability and protect users from abuse. Fourth, the OpenClaw moment foreshadows a broader shift: expect more “agent-as-a-service” offerings that blur the line between software and services, with regulatory and platform friction likely to intensify.

In short, OpenClaw’s Beijing surge is not a niche curiosity. It’s a preview of a future where autonomous AI agents become a common service layer, bought and managed by a growing workforce of remote operators. For product teams, the takeaway isn’t to chase the next flashy capability—it’s to harden the processes, guardrails, and governance that will let these capabilities scale responsibly this quarter and beyond.