How Ropes & Gray is developing AI for transactional diligence

Law firm Ropes & Gray says it is building a proprietary transactional diligence tool with OpenAI. Before an acquisition closes, buyers have to verify the seller's legal and financial claims, which usually means teams of lawyers working through enormous document sets on short deadlines.

The tool takes a first pass: it inventories and categorizes a data room, groups related documents by theme and issue, writes summaries, and drafts a report in the firm's own format. The firm has publicly described an ambition to produce a detailed, issue-level report in hours rather than weeks. If that holds, the economics and pace of deal work change.

This is a use of AI we find easy to defend. Diligence is high-volume reading against a checklist, the kind of work where consistency matters and fatigue causes mistakes. A partner at the firm put the value well: clients pay lawyers for judgment about the contracts, not for reviewing them. Pushing the grind to a machine and the judgment back to people is the right division of labor.

The design choices are encouraging. Ropes & Gray says it tested outputs against human-written diligence reports from completed deals and refined the rubric based on where the drafts fell short. Lawyers can inspect the groupings, and the firm says lawyers still review everything before it goes to a client. That's the right posture for a field where an overlooked clause can cost a client real money.

Now the caution. This is a case study published by OpenAI about its own partner, so we read it as a description of intentions and process, not independent evidence of accuracy. We don't see error rates, deal counts, or time savings measured on live matters. And the hours-not-weeks goal is described as an ambition, not a result.

The bigger issue is the career ladder. The firm compares the tool to the work of a capable junior-to-mid-level associate. That's a telling benchmark. Junior associates traditionally learn the craft by doing exactly this kind of first-pass review. If a machine does the first pass, firms need a deliberate answer to how new lawyers build the instincts to catch what the machine misses. Without one, the profession could save time now and hollow out its own pipeline of senior judgment later.

There's also a billing question underneath it all. If diligence that used to take weeks takes hours, either clients pay less or firms find new ways to price value. I'm watching which way that goes, because it'll tell us whether AI productivity in law reaches clients or stays with firms.

Ropes & Gray describes itself as an early AI adopter in the legal industry. The initial work has focused on material commercial contracts. The firm has 340 partners and 1,500 lawyers, and its head of practice AI, Howard Glazer, says the next stage is using AI to build collective knowledge across the firm, drawing on client meetings and past projects.