How Should Law Firms Change in the Age of Generative AI?

Hello, I'm Noriaki Asato, Representative Attorney at LegalAgent.

The process of engaging a law firm involves internal work that is hard to see from the outside. Before making a request, the in-house legal team reads through the consultation from the business division, organizes the issues, looks for related materials, and then drafts the questions for the attorney. After the answer arrives, the team has to break the legal opinion down for the business division and translate it into an actual negotiating strategy. Looking only at the time spent consulting outside attorneys does not reveal how heavy the coordination burden borne by the in-house legal team is.

Law firms, for their part, have long organized their work on the assumption of manpower. Partners, associates and staff review the work in turn before an answer is sent back. In large matters, this depth is sometimes necessary, but in day-to-day corporate legal work, as the back-and-forth of reviews increases, answers tend to take longer and fees tend to grow. Even when the business division wants to move a deal forward quickly, a situation arises in which only the legal review moves on a different timeline.

The spread of generative AI is an opportunity to rethink both of these problems at the same time. There is room for a law firm to go beyond simply being a firm that has AI tools, and to become a firm that enters directly into a company's workflow and supports decision-making at a distance close to that of an in-house legal team. This is less a question of choosing tools than an issue concerning the organizational design of the law firm itself.

The Structure of Traditional Law Firms and the Problem of Business Speed

Law firms have pyramid-shaped organizations in which work is reviewed in layers according to role and experience. A matter moves up in turn from staff to associates to partners, and comes back down after review. While this has value for training junior attorneys and for quality control, the increase in waiting for reviews can delay answers.

In work that requires prompt decisions, such as checking contracts or day-to-day legal consultations, delayed answers directly affect the progress of the business. The business division is pressed for a response in the middle of a deal, and in fundraising, discussions with investors, commercial registration and dealings with existing shareholders proceed in parallel. When launching a new business, too, product development and sales decisions tend to stall while waiting for legal's view. In corporate legal work, an answer may not be usable by the business division unless, in addition to being legally accurate, it arrives in time for the business deadline.

There is a limit to how far productivity can be raised simply by adding people. Even if you increase the number of attorneys, the process itself, taking in the request, reviewing materials, and researching and drafting, does not change.

Dividing Initial Work by Generative AI from Attorney Judgment

Picking out the issues to be considered from a contract, drafting proposed clauses and comments explaining the reasons for revisions, reading the counterparty's intent behind its revisions and putting together a response, and referring to similar past cases to show what differs this time. These are tasks that associates have traditionally spent time on, and they are also tasks for which drafts can be left to generative AI.

Of course, AI does not become an attorney. The attorney must personally take responsibility for judgments about the soundness of legal conclusions, negotiating strategy, business priorities, and whether something can be presented externally with responsibility.

However, once AI can perform the initial issue-spotting and drafting with high accuracy, the way attorneys use their time changes significantly. Less time is spent writing from scratch, and attorneys can concentrate on considering what should be revised, which risks should be communicated to the company, and which issues should be left to business judgment. Because the quality of AI drafts varies with the work and the materials, the role a law firm should play is precisely to check the accuracy of the quickly prepared draft and turn it into an answer that fits the company's particular circumstances.

I am thinking of an "inverted pyramid" structure in which initial work is left to AI and experienced attorneys make the judgments. Instead of many people working in stages, AI handles the initial work, and an experienced attorney makes quick judgments based on that output and finalizes the answer.

In reviewing an investment agreement, AI first extracts the issues and presents candidate revisions from both the investor's and the company's perspectives. The attorney then decides which clauses should actually be revised, taking into account the progress of the fundraising, the shareholder structure, and the impact on the next round. For a service agreement, AI extracts issues such as the acceptance procedure, ownership of intellectual property, and damages, and the attorney checks for omissions and errors. On that basis, the attorney decides how hard to negotiate by assessing the company's background, including whether there is an ongoing relationship with the counterparty, the profit margin of the deal, the delivery deadline, and the bargaining power of the business division.

Taking On the Preparation Before Consultation and the Work After the Answer

The aim of using technology at a law firm is not simply to speed up writing. It is to create a structure in which outside attorneys enter the company's workflow like members of the in-house legal team and support internal decision-making.

This is where the burden of pre-processing and post-processing mentioned at the beginning comes back in. Under the conventional framework, even though the company had engaged outside attorneys, a heavy coordination burden remained with the in-house legal team, and in reality the in-house team had to act as a sophisticated coordinator in order to make use of the outside attorneys.

By using AI, it becomes possible to consider having outside attorneys take on the preparation before the consultation and the work after the answer as well. The outside attorney directly reviews rough requests from the business division, chat exchanges, the history of past contracts and so on, identifies the issues, follows up with the business division, and presents options in line with the company's internal decision criteria. By going this far, the outside attorney changes from someone who answers questions into someone who moves internal decision-making forward. If you can entrust not only the outsourcing of routine work but also the sharing of internal decision criteria and coordination with relevant departments, you can reduce the workload carried by the in-house legal team itself.

A Legal Structure That Keeps Fixed Costs Down and Uses Variable Costs

For a company, hiring dedicated legal staff is a fixed cost. Bringing in talented people can be a great strength, but the expertise and volume of work a company needs change over time.

For a startup, investment agreements, shareholders' meetings, registration applications and dealings with existing shareholders are concentrated in fundraising periods. When launching a new business, drafting terms of use and a privacy policy and checking the relevant laws all come at once. In quieter periods, on the other hand, routine contract checks are the main work. If you try to absorb these fluctuations in workload with in-house employment alone, you tend to end up short-staffed in busy periods and weighed down by fixed costs in normal times.

Another approach is to use outside legal services in which AI and attorneys collaborate, matched to the workload in each period. Bringing in a practical legal function as a variable cost before adding headcount as a fixed cost is a realistic option for growing companies. However, simply piling up one-off legal consultations does not produce something that works like an in-house legal team. It takes hold as a legal function that is a variable cost yet has continuity only once you have built a structure in which, through ongoing involvement, the company's own decision criteria and past history are accumulated and attorneys who understand the circumstances handle the work.

Work Operations Rooted in the Actual Working Environment

For a law firm to provide sophisticated support consistently, it needs to redesign its work operations themselves rather than relying only on what is in individual attorneys' heads.

Taking contract review as an example, at intake we confirm the type of contract, the client's position and the desired delivery date. AI reads the contract and identifies the key clauses, potential risks and questions to confirm. The attorney organizes these into proposed revisions, comments for the counterparty and a risk explanation for internal use in light of the company's decision criteria, and finally delivers them in the form of tracked changes and comments in Word. With this process in place, the requester's experience comes closer to handing a matter to the in-house legal team than to having had a legal consultation.

If the delivery format does not fit actual practice, no matter how high-quality the answer is, the in-house legal team's burden does not decrease. That is because the person in charge is left with the work of deleting comments, rewriting expressions and recreating the explanation for the business division based on the answer received.

It also matters whether AI is embedded in the everyday working environment. In legal work, people keep tracked changes in Word, add comments, and work while checking against the materials in the matter folder. With a tool that only gives you answers in a browser, separate work remains if you are to keep using it in daily practice. I believe that only when AI is built into the whole sequence of work, from taking in the request to extracting issues, attorney review and editing in Word, does it lead to shorter working time.

At LegalAgent, through collaboration between our AI agent and attorneys, we deliver contract review, legal consultations, M&A work and more in a form that is easy for companies to use on the ground. If you want to speed up your legal work, rethink how you use outside law firms, or use a sophisticated legal function while keeping fixed costs down, please feel free to contact us.

Details of our legal outsourcing services are provided on the following page.

• Legal Outsourcing