Crosby launches Crosby Intelligence research arm and RedlineBench

On 17 June 2026, Crosby, the AI-powered law firm for contracts, launched Crosby Intelligence, a new research arm focused on advancing legal AI and building "agentic attorneys" for complex legal workflows. The initiative reflects the company's view that law is one of the next major frontiers for AI, particularly in areas where judgment and decision-making matter more than factual recall. Staffed by researchers, applied AI engineers and technical lawyers, Crosby Intelligence aims to publish benchmarks for legal domains where judgment matters most, simulate negotiations, and move time-to-signature from weeks to hours.

RedlineBench, a multi-turn contract-negotiation benchmark built with micro1, is the initiative's first major output. It measures how frontier models handle the full workflow of senior commercial lawyers across several rounds of live negotiation rather than isolated clause edits. On the firm's initial results, OpenAI's ChatGPT 5.5 scored highest at 50.5%, ahead of Google's Gemini 3.5 Flash at 45.1% and Anthropic's Claude Opus 4.8 at 44.4%. Crosby noted that human lawyers still outperformed models at finding new routes to resolution, while AI tools tended to anchor to their initial positions.

Over the coming months, Crosby Intelligence plans to keep publishing benchmarks on the field's hardest open problems, host a monthly speaker series, and — in conjunction with OpenAI — fund two research fellows. New York-based Crosby has raised over $85M from backers including Sequoia Capital, Index Ventures, Lux Capital, Bain Capital Ventures and Elad Gil.

Read more: Artificial Lawyer · John Sarihan (X) · Crosby Intelligence