Chinese robotics companies are rapidly integrating the viral open source AI agent framework OpenClaw into physical machines, moving the technology from digital chat interfaces into real-world applications far faster than their Western counterparts.
At the Consumer Electronics Expo in Shanghai last week, Ecovacs unveiled Bajie, a new household robot powered by OpenClaw. The machine is marketed as a home butler capable of tidying shoes, putting away toys, and responding to natural language commands for basic household tasks.
Ecovacs founder Qian Dongqi told Chinese outlet Ifeng that the long-term goal is for robots like Bajie to handle a wider range of chores and eventually function as autonomous home companions.
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
A reporter from the Chinese tech outlet 36Kr, who tested Bajie in person, observed that the robot often required multiple prompts to complete tasks correctly and exhibited unstable situations, a common early-stage limitation of agentic systems operating in unstructured physical environments.
Beyond home robotics, developers have embedded OpenClaw into Unitree’s G1 humanoid robot, enabling it to interpret natural language commands and navigate physical spaces in real time. A U.S. based team called Dimensional has open-sourced the integration code, accelerating experimentation in the Chinese robotics community.
AgileX Robotics published a detailed guide earlier this month showing how OpenClaw can control its robotic arms through conversational prompts, allowing users to direct precise manipulation tasks without traditional programming. Xiaomi is also actively testing customized versions of OpenClaw across its ecosystem, from smartphones to smart home devices, as part of its broader MiMo V2 Pro agentic AI initiative.
China has experienced a widespread OpenClaw adoption wave in recent weeks. Users have rushed to install the agent on personal devices, with some paying strangers for setup assistance and others forming long queues outside Tencent’s Shenzhen headquarters and Baidu’s Beijing offices to receive help from engineers. The viral phrase ‘raising the lobster’, slang for deploying OpenClaw to automate repetitive daily tasks, has become a cultural meme reflecting the tool’s rapid mainstream penetration.
In response to surging demand, China’s tech giants have moved quickly. Tencent, Alibaba, and ByteDance have each launched their own localized OpenClaw-inspired agent platforms in the past few weeks, integrating the open-source framework with domestic super apps such as WeChat, Alipay, and Douyin, as well as hardware ecosystems.
U.S. Security Concerns Mount
In contrast, U.S. tech leaders and researchers have raised escalating alarms about the security risks of granting agents broad system access. Meta’s alignment director, Summer Yue posted on X last month that after connecting OpenClaw to her inbox, the agent attempted to delete her emails without authorization.
“I had to run to my Mac mini like I was defusing a bomb,” she wrote.
The Information reported Thursday that another OpenClaw instance triggered a major internal security alert at Meta after acting without approval and exposing sensitive company and user data to unauthorized staff.
Elon Musk amplified the concerns on X, posting an image of a monkey being handed a rifle with the caption, ‘People giving OpenClaw root access to their entire life.’ Even Nvidia CEO Jensen Huang, who has praised OpenClaw as the most important software release, probably ever, has emphasized the urgent need for stronger safeguards.
Nvidia is developing its own agent platform, NemoClaw, with a heavy focus on enterprise-grade security, access controls, and behavioral monitoring.
The contrasting trajectories highlight a deepening divergence. China is racing to deploy OpenClaw and similar agents into physical robots and everyday workflows, leveraging open source accessibility and domestic super app ecosystems to achieve rapid scale. The U.S., meanwhile, is grappling with the security, alignment, and governance risks of giving agents broad system access, a tension that has slowed enterprise adoption while fueling public and regulatory scrutiny.
The Bajie robot and Unitree, as well as AgileX integrations, are seen as signs that agentic AI is moving beyond chat interfaces into embodied applications far faster in China, where regulatory and commercial incentives favor rapid experimentation.