How a Beijing Dumpling Chef Turned His Restaurant Into an AI Agent Test Lab

A Dumpling Shop That Talks to Your Phone

Li Bo runs a dumpling restaurant near one of Beijing's top technical universities, but his mind is fixed on what he calls "the most advanced productive force." In April, the 41-year-old restaurateur built a custom AI skill that lets diners check menus, ask for recommendations, or join the queue simply by talking to their personal AI agents - software that can execute tasks across a phone or laptop without manual taps or clicks.

The skill is a small file that teaches an AI agent a specific workflow. In this case, it teaches an agent how to interact with Jinguyuan, Li's two-branch dumpling operation. The move earned him headlines across Chinese tech media, not because the feature is driving foot traffic - Li freely admits it isn't - but because it signals something larger: a belief, widely shared in China's tech and business circles, that agentic AI will soon mediate everyday commerce.

Li holds a master's degree in telecommunications engineering. After graduating in 2010, he chose a different path from his classmates, working in hostels and selling homemade desserts before teaming up with a dumpling chef to launch Jinguyuan. The restaurant serves students and tech workers in the university district, a neighborhood steeped in engineering optimism. When agentic AI tools began circulating in China earlier this year, Li saw an opening to rejoin the technology conversation he had left behind more than a decade ago.

Why China's Small Businesses Race Toward AI

At DailyTechWire, we've tracked dozens of consumer AI launches across Asia over the past eighteen months, but the velocity of adoption in China stands apart. A 2025 survey conducted by Ipsos across thirty countries found that Chinese respondents expressed the highest level of enthusiasm for artificial intelligence - higher than any other nation polled, including the United States and South Korea.

That enthusiasm is visible on the street. Crash courses on large language models proliferate online. Office workers experiment with new AI tools to avoid falling behind in a fiercely competitive labor market. When the agentic tool OpenClaw went viral in China, people lined up to have it installed on their laptops, treating it less like a novelty and more like infrastructure.

The contrast with Western markets is sharp. In many developed economies, AI adoption among small businesses is quiet, sometimes furtive. In New York, menus and storefront posters bear the unmistakable fingerprints of AI-generated imagery and typography, yet few proprietors advertise their reliance on the technology. There is a cultural undertone of embarrassment, a sense that leaning on AI is akin to cheating or cutting corners.

In China, that stigma is largely absent. Worries about job displacement and the exploitative practices of big tech platforms are real, but they coexist with a pragmatic acceptance that AI is an inevitable force - one that individuals and businesses must ride rather than resist. Li's decision to build an AI skill for a dumpling shop, knowing it would generate more press than revenue, reflects that calculus. He sees himself as a role model, demonstrating that even small, traditional businesses can participate in the next wave of technological change.

Building in Public, Even When Usage Lags

Beyond the AI skill, Li also built a website that tracks real-time queue status at Jinguyuan's two locations. He wrote most of the code using Qoder, Alibaba's coding assistant, in a process he describes as "vibe-coding" - a term that has gained currency in China's developer community to describe collaborative work with AI where the human sets direction and the model fills in the syntax.

In July, Li struck a partnership with an AI cloud provider. Every customer who dines at Jinguyuan now receives a token coupon worth 10 yuan (roughly $1.50) that can be applied toward subscriptions to AI model services. The promotion is less about driving repeat visits and more about evangelism, encouraging diners to experiment with the same tools Li uses to run his business.

Li is candid about the results. The AI skill has brought far more media attention than actual usage. Most customers still order in person or through conventional delivery apps. But Li frames this as a long game. He invokes a historical parallel: when steam engines first appeared, skeptics mocked them as inferior to horses. Over time, internal combustion won out. Li believes the same arc will play out with agentic AI, and he wants his restaurant to be part of the early adoption curve, even if the immediate payoff is negligible.

The Deng Xiaoping Doctrine, Applied to Dumpling Sales

Li's framing of AI adoption is steeped in the language of Chinese modernization. He quotes Deng Xiaoping's famous dictum that "science and technology are a primary productive force," a line that underpinned China's reform and opening in 1978 and the explosive growth that followed. For Li, surrounding himself with the most advanced productive force is not a marketing gimmick; it is an article of faith.

This ideological backdrop helps explain why AI adoption in China often feels less like a consumer trend and more like a national project. The government has signaled strong support for AI development, particularly in areas that enhance productivity and competitiveness. Local governments offer subsidies and tax incentives for AI-related ventures. Universities emphasize AI literacy in their curricula. The result is an ecosystem in which experimenting with AI - even in a dumpling shop - carries a sense of civic and economic duty.

Li's background in telecommunications engineering also shapes his approach. He understands the infrastructure layer beneath consumer-facing AI tools: the model APIs, the cloud compute, the latency trade-offs. That technical fluency allows him to build quickly and cheaply, using off-the-shelf tools like Qoder and partnering with cloud providers who see value in associating their brand with grassroots adoption stories.

What Happens When Agents Become Customers

The broader question Li is testing is what commerce looks like when the customer is not a human but an agent acting on a human's behalf. If a significant portion of daily transactions - ordering food, booking appointments, comparing prices - are delegated to AI agents, then businesses will need to make themselves legible to those agents. Skills, APIs, and structured data become the new storefront.

This shift is already underway in China's e-commerce sector, where platforms like Taobao and JD.com are experimenting with agent-friendly interfaces that allow bulk queries, automated negotiation, and dynamic pricing. Li's dumpling shop is a microcosm of that shift, a single data point in a much larger transformation.

There are risks. If agents mediate most transactions, the businesses that win will be those with the best agent integrations, the cleanest data, and the strongest partnerships with the platforms that host those agents. Small operators like Li could find themselves locked out if they lack the technical resources to keep up. The same tools that promise democratization could also concentrate power in the hands of a few platform giants.

For now, Li is betting that the window for experimentation is still open. He is building in public, sharing his code and his thinking, hoping that other small business owners will follow. His restaurant may not see a surge in AI-driven orders this year, but he is positioning himself - and by extension, his business - at the edge of a shift he believes will define the next decade.

The Role Model Economy

Li's story resonates because it mirrors a broader pattern in China's innovation culture: the willingness to be a visible early adopter, even at personal or financial cost, in order to signal alignment with the future. This dynamic is not unique to AI. It has played out in electric vehicles, mobile payments, and short-video commerce. In each case, early adopters became cultural icons, their choices validated not by immediate returns but by their position in a narrative of progress.

Li frames his AI adoption in these terms. "Someone needs to start using it, even when it's not that good," he says. The implication is that early adoption is a form of leadership, a way of pulling the future forward. This mindset is less common in markets where skepticism toward new technology is more socially acceptable, where waiting for proven results is seen as prudent rather than passive.

At DailyTechWire, we've observed that this role-model economy creates its own momentum. When a dumpling shop owner becomes a media fixture for building an AI skill, other small business owners take note. The threshold for trying similar experiments drops. The cultural permission to invest time and resources in unproven technology expands. Whether this leads to sustainable business outcomes is still an open question, but it undeniably accelerates the pace of diffusion.

Li's dumpling shop is unlikely to change the trajectory of AI development on its own. But as a signal of how a segment of Chinese society is orienting itself toward agentic AI - pragmatically, optimistically, and with a sense of historical inevitability - it offers a window into the forces shaping adoption patterns across the region. The steam engine analogy may be overused, but in Li's telling, it carries weight. He is placing a bet that the future of commerce will run through agents, and he wants to be ready when it arrives.