Deal Velocity, Not Billable Hours: How Crosby Uses AI to Redefine Legal Contracting

Ryan Daniels and John Sarihan of Crosby explain why building an actual AI law firm — not legal software — enables the feedback loops and liability structures needed to automate contract negotiations at scale.

• Crosby charges per document, not by the hour; median contract turnaround is under one hour targeting minutes.
• Clients named include Cursor, Clay, and Unifi — fast-growing B2B companies signing dozens of contracts per day.
• Foundation models plateau at ~90% accuracy on legal tasks; getting to 99%+ requires per-customer fine-tuning and evals.
• Key north-star metrics are TTA (total turnaround time across all negotiation turns) and HURT (human review time).
• The billable hour only became standard in the 1950s; Crosby treats its demise as obvious, not controversial.
• Vision: dual AI agents simulate full contract negotiations between parties, producing an auditable back-and-forth record before humans review.
• The 92% of US law firms serving individuals (leases, child support) represent net-new automation opportunity — not lawyer displacement, since few serve those clients today.
• In-house legal teams grew 200% from 2007–2017 vs. 30% for law firms; Ryan sees AI-first specialized legal companies accelerating that shift.

2025-09-02 · Watch on YouTube