Incentive Mismatch in Legal AI
Scott Stevenson, CEO of Spellbook, on building Cursor for contracts
This incentive mismatch is why legal AI is landing fastest where lawyers get paid for outcomes, not hours. A contract copilot that turns a two day redline into 20 minutes creates obvious value for in house teams and flat fee firms, but inside big law it can directly shrink the inventory of billable time. That helps explain why adoption is strongest in transactional workflows with clear turnaround pressure and cleaner ROI.
-
Spellbook is built around the contract workflow inside Microsoft Word, where a lawyer drafts, reviews, and redlines agreements. That makes the value proposition simple, finish the same contract faster. This fits in house legal teams, procurement, and sales, where faster approvals move revenue and reduce bottlenecks.
-
Large firm buying behavior often separates purchase from usage. Tools can be approved by innovation committees and announced to clients, yet real day to day adoption stays muted because the people doing the work are still rewarded for spending more time, not less time, on the document.
-
The market is splitting by customer economics. Harvey has scaled by selling broadly into large firms and enterprises, while workflow specific tools like Spellbook, Luminance, and Ironclad win where contracts are already treated as an operating process to speed up, standardize, and close, not a time unit to bill against.
The next phase of legal AI will favor products tied to fixed fee work, in house operations, and business teams that live in contracts every day. As more routine drafting and review compress toward software speed, law firms will shift from selling lawyer hours to selling judgment, responsiveness, and packaged legal outcomes.