This early winter, during a break on a trip home, I did something that felt especially meaningful to me: for the first time, I brought the legal-AI conversations I'd been hosting in Silicon Valley back to China — to Beijing.
On November 21, together with Professor Gao Xiqing, senior partners and AI-specialist lawyers from Han Kun Law Offices, core members of leading domestic legal-tech companies, and nearly 60 legal professionals, big-tech engineers, and founders from both Silicon Valley and Beijing, we held a closed-door session of remarkably high information density.
Many friends in the room had taken time off to come; some traveled in from other cities. A few scholars and founders from Stanford joined online, waving at us through the screen. Looking at all those faces — some familiar, some new — I suddenly felt: the things I've been doing are actually starting to connect into a line.
From Silicon Valley to Beijing: Breaking Through the Wall
Two months ago, as a Silicon Valley lawyer acting purely on personal curiosity, I organized the community's first in-person event in the Bay Area — a very small discussion at the time. As of today, we've connected more than 600 lawyers, in-house counsel, algorithm engineers, and founders interested in legal AI. Most are spread across Silicon Valley, New York, and Washington, with others in the UK, Canada, Australia, and Beijing, Shanghai, and Shenzhen. Our discussion last month had just wrapped up on the Stanford campus — a conversation with members from leading US legal-tech companies including Harvey, GC AI, and Advocacy.
Through a chance encounter, I took part as a guest in a legal-AI discussion that Han Kun hosted in Silicon Valley. There, I realized that many of the concerns and questions domestic lawyers have about AI overlap with the topics we discuss in the Valley. Together, we made this Beijing closed-door session happen.
Gao Xiqing: Lawyers Need the Wisdom to Move Through Cycles
"The tools change; the lawyer's core does not." As our featured guest, Professor Gao Xiqing — former Tsinghua University Cheng Yu-tung Chair Professor of Law — set a calm, pragmatic tone for the event. Professor Gao was one of the key architects of China's early capital-market institutions, an influential scholar in US–China legal and financial exchange, and a longtime mentor whose encouragement has shaped me personally.
Facing the young lawyers in the audience, Professor Gao told a story from several decades ago. As a student at Duke Law School, he worked part-time in the library and so became intimately familiar with the paper-based legal-research system of the day — an era when you had to "Shepardize," peeling back layer after layer to find the relevant cases.
Using AI, he argued, is simply the Shepardize of a new era. From paper volumes to databases to today's large models, the form of the tool keeps evolving — but the lawyer's core skill of reasoning step by step, unraveling a problem thread by thread, has never gone out of date. He cautioned young lawyers not to assume too readily that "entry-level skills will be entirely replaced by AI." Every mature lawyer must go through foundational training; no one gets to skip it.
He closed with a line many in the room remembered: "Decades from now, when you look back, you may marvel that you caught a great wave — or you may feel it was all just a small joke at the time. You're standing in a very unique position. The road ahead is long. Don't let yourself have regrets."
Han Kun's Playbook: "Handle With Care, Move Fast in Small Steps"
When everyone is talking about AI, how does Han Kun — a representative of China's top-tier firms — actually view it?
In his opening remarks, Han Kun partner and co-head of the corporate compliance practice Ma Chen framed three kinds of dialogue in the AI era: the role-based dialogue between the two sides of a deal, the mutual learning between US and Chinese technology paths, and the interplay between technology and the humanities.
Ma Jiao, the partner in charge of AI within Han Kun's knowledge-management department, summed up the firm's strategy in a single phrase: "Embrace it actively, handle it with care, move fast in small steps."
The bind many firms face today is this: they want to use AI, but don't dare put their data on the cloud; they want to build systems, but fear investing tens of millions only to see it become obsolete overnight. Han Kun's answer is highly pragmatic — "private deployment + micro-tool innovation."
- Data is your confidence; security is your bottom line. Han Kun began its knowledge-management (KM) work back when the firm was founded in 2004, and that accumulation became the most solid foundation for understanding and applying AI. Putting data security first, they adopted private AI deployment so the technology could do real work within controlled boundaries.
- Start from pain points; build small tools. Rather than chasing something big and all-encompassing, they built small tools like an automatic signature-page generator. It solves just one tedious pain point in a transaction — but it genuinely frees up lawyers' energy.
She was candid that today's legal-AI products still fall well short of a firm's standards — AI still "hallucinates" a fair amount, and struggles with drafting and reviewing complex bilingual (Chinese–English) investment contracts. But in the long run, AI will replace the lawyers who don't use AI, and mapping out business processes will become the prerequisite for a lawyer to use AI tools well.
Huayu Yuandian: A Vertical Breakout Under General-Purpose Large Models
Chang Cheng, head of legal-SaaS product operations at Huayu Yuandian, spoke from the technology-supply side, revealing the "dimensional strike" — and the new opportunities — that legal-tech companies now face.
- The common predicament for legal tech in the large-model era. He noted that AI used to rely mainly on meticulous manual training, whereas in the large-model era, AI can "self-discover" capability boundaries across many scenarios simply by changing the "input." Because general models improve so fast, vertical tech companies often just build a "shell" over industry data — with two results. On one hand, domestic legal-tech products are severely homogenized, clustered in contract review, similar-case retrieval, and text generation. On the other hand, the products' added value becomes a mystery — users ask, "Compared with a general-purpose model, what exactly are you better at?"
- Where's the breakthrough for legal AI? On applications, the focus should be bottom-up exploration in non-consensus areas. He cited niche tracks that some lawyers had built themselves — a "securities-loss calculation tool," a "patent-application assistant," a "creditor dialogue bot for bankruptcy cases" — as well as extensive use of AI in client-acquisition scenarios. On product design, the key is lowering the user's cost of verification: why do programmers love AI so much? Because you run the code and immediately know if it's right, whereas a lawyer using AI to generate a contract has to check it word by word — the verification cost is too high. As the first company in China to build judicial-informatization projects, what Huayu Yuandian is doing is exactly this: using technical means (like one-click statute tracing and validity flagging) to make AI genuinely "verifiable and usable."
Finally, he raised a deeper point: lawyers' resistance to AI is, to a large extent, a matter of perception. The industry keeps agonizing over "can AI replace lawyers, can it be zero-hallucination," while overlooking a more important question — "who exactly can AI help, at what step, and to what degree?" Only when the industry starts from value rather than fear can the technology's incremental gains become visible.
Roundtable: AI's Impact, US–China Trends, and the Future of the Legal Profession
After the keynotes, I invited Li Rong, VP of Power Law Intelligence, and Silicon Valley legal-tech founder Shaoting Qin to join our guests for a lively roundtable on three themes: "how AI is reshaping legal services," "US–China legal-tech trends and collaboration," and "career development for legal professionals."
Along the way we also touched on the practical shape of the "AI law firm," how leaders like Harvey handle clients' data-security concerns, the differences between US and Chinese courts' "judicial AI models," and the explosive rise of emerging roles like "Legal Engineer" in the US legal market. Given the length, we'll do a separate write-up of the roundtable — stay tuned.
In the final open-discussion segment, audience members — including a legal evaluator for ByteDance's Doubao and a founder of a compliance-focused legal-tech startup — raised their hands to share cases. The discussion ran so hot that we went an hour over time. Thank you to everyone who stayed to the very end.
Next stop — see you again.