Spellbook building always on contract AI
Scott Stevenson, CEO of Spellbook, on building Cursor for contracts
This points to the real prize in contract AI, moving from a tool a lawyer opens on demand to a system that watches the contract estate all day and starts work before anyone asks. For Spellbook, that means expanding from Word based redlining into intake, routing, storage, and ongoing monitoring, so the product can notice a law change, find the affected agreements, draft the first pass fixes, and hand legal a queue of decisions instead of a blank page.
-
Spellbook already sits in the drafting loop where lawyers spend time, inside Microsoft Word, altering contracts with track changes against a customer playbook. That matters because proactive agents only work if they are plugged into the place contracts are created and revised, not just a separate chat window.
-
The next step is closer to AI paralegal work than chatbot work. The roadmap described is automated intake and triage from email and Slack, first pass review before a lawyer opens the file, and risk surfacing across large contract portfolios. That turns contract software from a repository into an operating layer.
-
This is also where the market is splitting. Ironclad and Icertis grew by selling configurable systems of record for enterprise contracts, while newer players like Luminance push autonomous review and negotiation. Spellbook is trying to combine the low friction Word native wedge with the always on workflow ownership that made CLM vendors durable.
The category is heading toward software that owns the full contract loop, from inbox intake to signed archive to post signature risk detection. The winners will be the products that can act in the background with enough context and trust to move work forward, while still fitting the daily habits of legal, procurement, and sales teams.