This afternoon, a friend I had not seen for a long time came to Mankun for a chat.
The last time we met was at the end of last year. She asked me whether I had been working on AI or continuing with blockchain over the past six months. I said that while our core business remains intact, the task to which I have devoted the most time this year is using AI to rebuild Mankun from the ground up.
After saying this, I paused for a moment myself.
Mankun has been operating for five years. Over the past five years, we essentially rebuilt a law firm on Feishu (Lark): customer management, projects, knowledge bases, content, finance, and team collaboration were all moved online as much as possible. Now that AI has arrived, I increasingly feel that our previous digitalization efforts were merely laying the foundation. Many workflows that have been running stably are worth deconstructing and re-examining.
My current understanding of an AI-driven law firm is one that redesigns business processes, organizational costs, and client services segment by segment.
Therefore, I recently felt as though I had returned to an internet company. Previously, I attended meetings with product managers and project managers; now, I act as the product manager myself, explaining business processes to AI, identifying bottlenecks, specifying requirements, and having it produce solutions.
This endeavor is both exciting and exhausting. My Codex subscription increased from $20 per month to $100, and then to $200. Often, I am ready to leave work only when my daily quota is nearly exhausted.
Stop Playing with Tools
In the month or two following the Spring Festival, I experienced a significant period of AI anxiety.
New models, new products, and new use cases emerged every day. One day someone would recommend this tool, and the next day someone else would deploy that one. During that period, I spent considerable energy installing, debugging, and testing various tools, fearing I might miss out on something important. However, after all this effort, I suddenly realized I had gone astray: I was playing with AI rather than truly integrating it into Mankun’s operations.
No matter how many tools one tries, they have no direct bearing on law firm management. An agent with an impressive demo remains merely a sophisticated toy if it cannot be integrated into client services, project management, content operations, and daily decision-making.
Subsequently, my approach became quite simple. Technology will continue to advance; problems that cannot be solved today may cease to be issues within two months. Rather than constantly switching among dozens of products, it is better to select a top-tier tool with sufficiently balanced capabilities, use it consistently and thoroughly, and refocus attention on business operations.
This shift was significant for me. From that moment on, I stopped chasing AI trends and began asking: Which tasks at Mankun are currently the most labor-intensive? Which processes are most prone to error? Which experiences recur frequently enough to be standardized into fixed capabilities? Which client needs have previously remained unmet due to prohibitive costs?
Whether AI is relevant to law firms is not answered at press conferences, but in these questions.
AI Must Enter the Workflow
My job involves sitting down with responsible attorneys to review business processes from start to finish: how things are currently done, where the bottlenecks lie, which judgments must be made by humans, and which actions can be automated by AI.
This closely resembles the work of product managers I observed in internet companies. The difference is that whereas previously one had to wait for R&D scheduling after submitting requirements, now I can handle many tasks directly with AI. A law firm does not need to maintain a full engineering team before it qualifies to transform its own workflows.
Service proposals and quotations provide another intuitive example.
After discussing with commercial clients, they typically require a service proposal tailored to their specific project circumstances. Previously, it was not uncommon for paralegals to spend an entire day conducting research, organizing information, writing, and formatting. Now, after the meeting recording ends, AI can quickly generate a fairly complete initial draft aligned with the client’s business context, issues, and timeline requirements, including formatting. The lawyer’s role is to determine whether the scope of service is correct, whether any risks have been overlooked, and whether the quotation aligns with the project’s realities, rather than spending half a day moving paragraphs and adjusting font sizes.
Changes in content operations are even more pronounced. At Mankun Law Firm, topic discovery, news aggregation, preliminary article review, SEO and GEO optimization, formatting, and multi-platform distribution are gradually being incorporated into automated workflows. If content operation roles continue to function in the traditional manner, their necessity becomes difficult to justify. As a manager, I cannot pretend that cost structures remain unchanged while observing that work methods have already shifted.
Interestingly, AI has not made me less busy; rather, it has made me busier. Previously, many ideas were constrained by manpower, budget, and development cycles, rendering them impossible to implement. Now, whenever I identify a business bottleneck, I am compelled to pursue further improvements. The time saved is quickly filled by new experiments.
The first wave of change brought by AI may involve workforce reduction, but a deeper transformation lies in enabling professional service firms, which previously lacked product development capabilities, to suddenly possess the potential to create products.
AI in the Front, Lawyers in the Back
Some time ago, I integrated the capabilities of a certain securities platform into my AI workspace.
Previously, to check my account, I had to open the app and manually review holdings, market data, and news. Now, AI can monitor abnormal fluctuations at scheduled intervals, generate daily investment reports after market close, and conduct weekly reviews combined with transaction records.
This small matter profoundly impacted me.
Translating this to legal services, in the future, clients may not need to locate a specific lawyer and recount their issues from scratch each time. Instead, they can first enter Mankun’s AI service portal. This portal understands the client’s projects, historical documents, confirmed conclusions, and current progress, and can leverage Mankun’s accumulated professional expertise. Routine consultations, preliminary contract reviews, document searches, and process reminders are initially handled by AI; lawyers step in when legal judgment, liability confirmation, negotiation, and complex communications are required.
I am increasingly inclined toward the view that legal services five years from now will likely feature AI in the front and human lawyers in the back.
This does not mean clients will be left to a public chatbot. On the contrary, clients need a service portal tied to specific projects, verifiable by lawyers, and aware of permission boundaries. AI can handle a large volume of repetitive and basic issues, but professional responsibility must still be borne by lawyers.
It is not merely a client acquisition tool.
AI on the client acquisition side and AI on the delivery side operate under different logics. The former helps clients identify problems and understand services, while the latter enters real projects and collaborates with human lawyers to complete work. What clients pay for is no longer just a few consultations or documents, but a continuously online human-machine collaborative service.
If this path proves viable, law firms in the future will sell not only lawyer hours but also productized knowledge, processes, and response capabilities.
AI Reshapes Law Firm Operations
Many people enjoy debating whether AI will replace lawyers. As the operator of a law firm, I am more concerned with a specific question: How will AI rewrite the cost and revenue structures of law firms?
Traditional law firms often equate scale with office space and headcount. Larger premises, more workstations, and more comprehensive middle and back offices appear indicative of a thriving institution. But who ultimately bears these costs? What value do they add for clients? If lawyers rarely come to the office, why should the law firm pay for “seats filled to look bustling”?
Both Mankun’s Shanghai and Shenzhen offices have recently moved to better premises, but one point remains consistent: each has only 20 workstations.
If all colleagues in the Shenzhen office were to work offline simultaneously, there would theoretically not be enough space for everyone. However, this poses no actual problem, as many lawyers already work online, and many do not require fixed workstations.
Cost reduction is certainly not the ultimate goal. By lowering fixed costs, a law firm creates room to invest money and energy in more valuable areas: professional branding, public case sources, client services, global collaboration, and lawyer teams that genuinely generate revenue.
Mankun does not currently strive to become a large-scale platform firm with numerous personnel. We prefer to establish a sufficiently clear brand in new economy sectors such as Web3, AI, and tech finance, thereby attracting lawyers who align with these directions and possess both professional expertise and revenue-generating capabilities. The firm’s content, marketing, and public case sources can support lawyers, while lawyers’ professional services further strengthen the brand. This cycle is more compelling than the traditional law firm model reliant on selling workstations.
Exploring the Future of Legal Services
I am currently seeking a new direction: Is it possible to build an AI legal technology company directly serving enterprise clients?
At present, I have only two somewhat vague assessments. The clients will likely be B2B, and the market will likely be overseas.
Domestic legal service fees are too low, so enterprises may lack sufficient motivation to use AI to replace certain lawyer tasks. Some law firms I visited in Hong Kong charge high fees, yet their files and daily collaborations remain heavily dependent on paper and email. The Web3 enterprises we serve are, in essence, Chinese internet companies expanding overseas. New product opportunities may emerge around the corporate, payment, data, licensing, anti-money laundering, and dispute resolution issues these enterprises encounter across different jurisdictions.
Recently, I also engaged with a Hong Kong team researching the use of AI to help victims of asset fraud organize report materials and clues. Traditional legal services may only cover clients involving larger amounts. If AI can reduce the costs of materials and processes, individuals with lower loss amounts who previously could not afford lawyers may also gain access to usable professional support.
For young legal professionals, I have an increasingly clear recommendation: Opportunities belong not only to those who are most proficient in using AI tools, but to those who can understand a business segment and then transform it using AI.
A recent law school graduate cannot immediately reconstruct an entire law firm. However, they can join a team, spend two or three months understanding their assigned module, and then automate and standardize one segment of it. After completing one, they proceed to the next. Through this process, they learn not only law but also business operations, while developing the instincts of a product manager.
The ability to write prompts is not scarce. The capability to understand who is doing what, why bottlenecks occur, which steps can be delegated to machines, and which steps must remain under human responsibility will become increasingly valuable.
During the first two years of entrepreneurship, I enjoyed discussing new models with peers, but later realized that discussion alone was not necessarily effective. Now, I prefer to take direct action.
Mankun is our testing ground. Any ideas about the future law firm are first tested on ourselves; if useful, they are retained; if not, they are discarded. I do not know whether this path will ultimately succeed. However, I am certain that AI’s impact on law firms will not stop at drafting contracts, researching cases, and creating summaries. It will penetrate cost structures, staffing, client relationships, service pricing, and brand organization, forcing every operator to reconsider: Why does a law firm need to exist in its current form?
Over the past five years, we built Mankun once.
Next, I intend to rebuild it using AI.
