Ropes & Gray builds AI tool with OpenAI for transactional diligence

A first pass for commercial contract diligence

Before an acquisition is finalized, the buyer has to investigate the seller's financial and legal claims. According to OpenAI's account, this transactional diligence uncovers risks and liabilities and usually requires an entire team of lawyers. Parsing an enormous number of documents and analyzing patterns can mean days of intensive work when a bid has to be assembled on a short timeline.

Ropes & Gray, which OpenAI describes as an early adopter of AI in the legal industry, built the tool in partnership with OpenAI. The tool combs through thousands of pages, identifies relevant documents and groups them automatically by themes and issues. It also produces summaries and a completed report in the firm's signature format.

The tool inventories and categorizes a data room, so it can review full relationships rather than isolated documents. Lawyers can inspect the grouping and categorization to confirm nothing is missed.

Howard Glazer, Partner and Head of Practice AI at Ropes & Gray, put the goal this way: "Our clients don't pay us to review contracts. They pay us to make judgments based on what's in the contracts."

How the tool was developed and tested

Initial development work has focused on material commercial contracts. In working sessions, the firm's lawyers compared test outputs with human-authored diligence reports from completed deals. They then refined the rubric and report format based on what the drafts got right and where they fell short.

Ropes & Gray designed the tool around a consistent query and framework that mimics the steps its team would take, from initial reviews to summaries to higher-level reviews. The source compares the output to the work of a capable junior-to-mid-level associate, serving as a first pass through the process.

Human review remains central. Josh Lichtenstein, Partner and Head of ERISA Fiduciary Practice, said: "Our lawyers still need to review everything." He added that anything delivered to a client must be something the firm can stand behind 100%, which is why the tool lets lawyers easily review interim summaries and the actual documents.

• The tool inventories and categorizes a data room, grouping related documents by themes and issues.
• The workflow produces summaries and a report in Ropes & Gray's format.
• Lawyers review interim summaries and source documents before delivering work to clients.
• Testing compares draft outputs with lawyer-authored diligence reports to refine the rubric and report format.

Ambitions for complex transactions

Ropes & Gray has publicly described an ambition for the tool to produce a detailed, issue-level diligence report in hours rather than weeks. According to the source, with tools like its proprietary diligence tool and GPT-6 Astra, the firm has been able to solve client problems far faster than before.

Lichtenstein and Glazer also shared two lessons. The first is to experiment with AI for everything: Lichtenstein says he uses it from finding dress shoes to vibecoding games for his kids, and that watching how to handle its mistakes turned it from something he played with into a powerful professional tool. The second is to put clients first, with Glazer saying the focus is on how to use AI to benefit them.

What comes next

Glazer believes the next stage of AI at the firm will help create collective knowledge. With 340 partners and 1500 lawyers, Ropes & Gray sees value in giving everyone access to each person's knowledge, from client meetings to past projects, to better solve client problems.

For the legal market, the example shows a model in which AI handles the initial document review while lawyers spend more time reviewing and refining the work, leading, in the firm's words, to a more carefully crafted final judgment. The source does not give pricing, availability or performance figures for the tool.