How law firms should change in the generative AI era
Hello, this is Legal Agent.
A request to a law firm carries hidden work on both ends: before the request goes out, in-house legal has already read the business unit's question, framed the issues, and turned them into something a firm can answer; after the answer comes back, someone still has to translate it for the business and fold it into the negotiating strategy. At many companies, that surrounding work takes longer than the time outside counsel actually spends on the file.
The pyramid moves on a different clock than the business
Traditional firms are built around a pyramid: a matter moves up from staff to associate to partner and back down again. That structure has real value for training and quality control on large matters, but every extra round of internal review adds half a day or a day to the answer, and for day-to-day contract review and quick legal questions, that delay lands directly on the business. An answer that arrives late is not much more useful than one that is wrong.
Where generative AI fits: the associate's early-stage work
Spotting the issues in a contract, drafting clause language or a comment, reading a counterparty's rationale for a proposed change and preparing a response, and flagging what differs from a similar past deal: this is exactly the work a strong junior associate has traditionally spent hours on, and it is also where generative AI performs well. AI does not become the lawyer; legal conclusions, negotiating strategy and the judgment call of what can go out under the firm's name still belong to the lawyer. But once AI can produce a strong first draft quickly, the lawyer's time shifts from writing from scratch to deciding what to fix and what risk to flag.
An inverted pyramid
Instead of many people working in sequence, AI can handle the bulk of the early-stage work while an experienced lawyer sits on top, making the judgment calls quickly and polishing the output. For an investment agreement, AI surfaces the issues from both the investor's and the company's side, and the lawyer decides what actually needs to change given the round's progress and the cap table.
Absorbing the work on both sides of the request
The real opportunity for an AI-native firm is taking on the framing work before the request and the translation work after the answer, not just answering faster. Outside counsel who can read a rough business-unit message directly, spot the issues, ask a clarifying question back to the business unit, and present options aligned with the company's own standards moves from "person who answers questions" to "person who moves the company's decision forward," and that is where in-house legal's actual workload comes down.