A top-tier law firm just made contract review billable at scale—and showed every other knowledge worker what automation actually looks like when you're not worried about your job disappearing.
The Summary
- Ropes & Gray built an AI system with OpenAI that groups documents, generates summaries, and produces reports for M&A diligence—lawyers still review everything before client delivery
- The firm isn't replacing associates, they're compressing the timeline between "dump of 10,000 files" and "actionable legal advice"
- This is the billable hour's last stand: automate the grunt work, keep the judgment calls expensive
The Signal
Ropes & Gray, the kind of white-shoe firm that bills $1,500 an hour for partner time, just built an AI pipeline for transactional due diligence. The workflow is straightforward: AI reads thousands of contracts and corporate documents, groups them by type and risk category, generates summaries, spits out reports. Then lawyers review the output before anything touches a client. The firm developed this with OpenAI, meaning they're running some version of GPT models under the hood, likely fine-tuned on legal taxonomy and their own document standards.
This matters because M&A diligence is where young lawyers used to learn the business. You'd spend weeks in a data room, reading every contract in a target company's filing cabinet, flagging risks, building charts. Soul-crushing work, but you learned how deals are structured. Now that pattern recognition happens in minutes, not months. The question isn't whether AI can read contracts—it obviously can. The question is what happens to the training ground.
"The firm isn't eliminating lawyers. They're eliminating the part of lawyering that feels like data entry."
Ropes & Gray's approach is telling. They kept humans in the loop at the review stage, which means they can still bill for the work and maintain quality control. But they collapsed the timeline. A diligence process that used to take three associates six weeks might now take one senior lawyer three days of review time. The client gets faster answers. The firm gets higher margins. The associates get... fewer hours doing the thing that used to make them valuable.
Key shifts this reveals:
- Professional services firms can adopt AI without cannibalizing revenue if they reframe speed as the product, not hours
- The "review AI output" job is real, but it requires senior judgment—junior roles evaporate
- Legal AI isn't coming from legal tech startups, it's being built in-house by the firms with the most to lose
The Implication
If you work in professional services—law, consulting, accounting, financial analysis—this is your template. The firms that survive will build AI that makes their experts faster, not cheaper. They'll sell speed and certainty, not labor hours. Which means the juniors who used to learn by doing the tedious work now need a different on-ramp to expertise. Law schools aren't teaching "how to review AI-generated diligence reports" yet, but they should be. And if you're building tools for knowledge workers, the wedge isn't "replace the lawyer." It's "make the lawyer 10x faster and let them bill the same rate."