Building the Contract Operating System
Scott Stevenson, CEO of Spellbook, on building Cursor for contracts
This reveals that the real battleground in legal AI is shifting from lawyer headcount to contract volume. Spellbook is built around the people who move deals forward, not just the lawyers who approve them, which is why it fits large in house teams, procurement, and sales ops. In practice that means living inside Microsoft Word, automating first pass review, and expanding into the intake, workflow, and storage layers around contracts.
-
Spellbook already gets about 60% of revenue from corporate in house teams, and in house is growing 3x faster than its law firm segment. That matters because in house legal, procurement, and sales all share the same goal, get contracts signed faster, instead of protecting billable hours.
-
Harvey and Legora are organized around the lawyer seat. Harvey grew to $195M ARR by the end of 2025 selling legal research, document review, and drafting into law firms and in house teams, while Legora built a lawyer workspace tied to Word and document systems. Spellbook is narrower on workflow, but broader on who can use it.
-
The bigger upside is budget expansion into contract infrastructure. Spellbook is moving beyond redlining into automated intake, triage, Slack and email triggers, and background review for teams handling up to 100,000 contracts a year. That puts it closer to CLM territory, where incumbents like Ironclad are already at roughly $150M ARR.
The next phase is that contract AI stops being a drafting helper and becomes the operating system for commercial agreement flow. The winner will not just answer legal questions, it will quietly process incoming contracts, route them, mark them up, compare them to policy and market terms, and hand humans only the exceptions that matter.