CLM Did Not Take Off
Scott Stevenson, CEO of Spellbook, on building Cursor for contracts
The real opening in contract software is no longer storing signed PDFs, it is doing the work before and around signature with far less setup. Older CLM systems won by asking legal ops teams to model approval paths, fields, and repositories in advance. The new AI wave is shifting value toward tools that can read inbound contracts in Word, email, and Slack, flag issues, draft edits, and route work without months of manual configuration.
-
Ironclad still shows there is a large CLM business, with estimated ARR reaching $150M in January 2025 and about $200M by February 2026. But its core strength came from highly configurable pre-signature workflows, a model that is powerful once installed and harder to spread broadly because setup is heavy.
-
Icertis is even larger, with estimated ARR of $350M by August 2025, and it is strongest where procurement, ERP integration, and compliance matter most. That supports the idea that legacy CLM did land in big enterprise accounts, but often as an implementation led system of record rather than an everyday drafting tool people naturally adopt.
-
Spellbook is pushing from the opposite direction. It starts inside Microsoft Word where lawyers already redline contracts, then expands toward intake, triage, workflows, storage, and proactive review for teams handling up to 100,000 contracts a year. That makes it less a bolt on to CLM and more an attempt to rebuild CLM around AI assisted execution.
The category is heading toward AI native contract operations, where the winning product will combine repository, workflow, and drafting assistance in one system that fills in metadata and starts the work automatically. Incumbents have distribution and installed bases. New entrants have the chance to define the default workflow because they remove the manual setup tax that slowed CLM adoption the first time.