Sullivan & Cromwell brings legal expertise into agreement review

Sullivan & Cromwell has developed Agreement Analyzer, an AI-powered tool that brings the firm's legal expertise directly into the contract review process. For lawyers at the firm, reviewing a draft agreement involves weighing terms against relevant precedents, applying approaches based on the firm's experience, and being mindful of client priorities. Agreement Analyzer integrates these inputs into a single workflow, producing analysis and proposed changes that lawyers can inspect and refine.

Partner John Savva originally built the tool as a single-user proof of concept using Codex, then collaborated with OpenAI to improve output quality, add parallel processing for efficiency, and enable simultaneous use by multiple lawyers. The tool combines Sullivan & Cromwell's workflows, curated firm precedents, negotiating playbooks, and ChatGPT Enterprise. A lawyer inputs a draft agreement along with selected precedents and a playbook containing firm and client positions and market practice information. The tool then performs multiple review cycles from different perspectives, synthesizes the outputs, identifies potential issues, and returns a memorandum with proposed tracked changes for the lawyer to review, adapt, or reject.

Savva emphasizes that transparency is essential for trust in AI output, noting that if a tool makes a claim about market practice, the source must be visible. He describes an adversarial design in which specialized agents review agreements in parallel for issues such as market deviations, risk allocation, clause interactions, and ambiguities, with other agents challenging their conclusions. Findings are distinguished as supported, dismissed, or unverified with qualifications. Sullivan & Cromwell continues to explore task-specific legal tools, and Savva points to advances in reasoning capabilities with GPT-6 Astra as a major step forward, addressing what had previously been a bottleneck.

(Source：OpenAI)