A small restaurant near a Beijing university has become an unlikely symbol of something much larger. Jinguyuan, a dumpling shop, built an AI “skill” that lets customers check the menu, get recommendations, or join the queue by talking to their personal AI agents. Its owner, 41-year-old Li Bo, is not a tech executive. He studied telecommunications engineering, then spent years running hostels and selling homemade desserts before opening a restaurant. That background makes his story more instructive, not less.
A Skill, a Queue Tracker, and a 10-Yuan Coupon
What Li actually built is worth understanding in concrete terms. An AI skill is a file that teaches an AI agent a specific workflow. In this case, the skill connects a customer’s personal AI agent to Jinguyuan’s systems, allowing the agent to act on the customer’s behalf: browse the menu, ask for suggestions, or secure a spot in line. Li built this in April, using his own judgment about where consumer behavior was heading.
He did not stop there. Li also used Alibaba’s coding tool Qoder to build a website that tracks queue status across Jinguyuan’s two branches, working primarily through a method known as vibe coding, where a user describes what they want in natural language and an AI tool generates the underlying code. In July, he partnered with an AI cloud provider to offer every customer a token coupon worth 10 yuan (approximately $1.50) that could be applied toward AI model subscriptions.
Li is candid about the results so far: the AI skill attracts more media attention than actual usage. He frames this not as a failure but as a necessary phase. His analogy is direct. Early steam engines, he notes, were mocked for being inferior to horses. The technology improved. The outcome is now history.
Excitement and Anxiety, Held at the Same Time
Li’s story sits inside a broader pattern that distinguishes China’s relationship with AI from that of many other developed economies. A 2025 survey by public opinion firm Ipsos, covering 30 countries, found that the Chinese population expressed the highest level of excitement about AI among all nations surveyed.
That enthusiasm coexists with real concern. People in China share worries about AI’s effect on employment and about the exploitative dynamics of the tech industry. The difference is not the absence of doubt. It is that doubt does not translate into hesitation. Office workers rush to experiment with new tools to stay competitive. When the agentic tool OpenClaw went viral, people lined up to have it installed on their laptops. Online crash courses on large language models have multiplied rapidly.
The contrast with other contexts is sharp. In places like New York City, small businesses rarely advertise their AI use openly, even when the visual evidence, AI-generated imagery on menus and posters, is visible everywhere. There is a cultural friction around AI adoption in many Western settings, something close to the discomfort associated with taking a shortcut. In China, the framing is closer to the opposite: not adopting AI is what feels like falling behind.
Li captures this with a quote he attributes to Deng Xiaoping, the leader whose reforms beginning in 1978 set off China’s decades-long economic transformation: “Science and technology are a primary productive force.” For Li, this is not nostalgia. It is a live operating principle.
What a Dumpling Shop Reveals About Technology Adoption
Here is what most coverage of this story misses. The significance of Jinguyuan is not that a restaurant built an AI feature. It is what the story illustrates about the conditions under which new technologies actually spread.
Adoption does not happen because a technology is mature. It happens because enough people decide to engage with it before it is mature, generating the feedback, the use cases, and the cultural familiarity that allow it to improve. Li knows his skill is not yet driving significant customer behavior. He is building it anyway, because he understands that the infrastructure of habit needs to be laid before the demand arrives.
AI agents, specifically, represent a meaningful shift in how software interacts with the world. Rather than a user opening an app and navigating its interface, an agent acts on the user’s behalf across multiple systems. The “skill” model that Jinguyuan adopted is one early attempt to make businesses legible to these agents. It is a primitive version of something that, if it scales, changes how discovery, recommendation, and transaction work at a fundamental level.
The broader question is not whether AI agents will become a significant part of daily life. It is which businesses, industries, and countries will have built the habits, the infrastructure, and the cultural readiness to participate when that shift accelerates.
In Short
A Beijing dumpling shop built an AI skill that lets customers interact with the restaurant through their personal AI agents. Its owner, Li Bo, is transparent that adoption is still limited, but frames early participation as the point, not the problem. China’s broader AI culture, marked by high enthusiasm even alongside real anxiety, creates conditions where this kind of experimentation happens outside the tech industry. The Ipsos survey finding that Chinese respondents expressed the highest AI excitement among 30 countries surveyed is not just a data point. It is context for understanding why a restaurant owner quotes Deng Xiaoping when explaining why he learned to code with AI.
Based on reporting from Rest of World.