Legal AI Augments, Not Replaces Lawyers
Scott Stevenson, CEO of Spellbook, on building Cursor for contracts
The winning legal AI products today make lawyers faster, not obsolete. Contracts are a strong AI use case because they are repetitive and standardized, but buyers still want a human who can explain the risk, bless the output, and set guardrails for sales and procurement teams. That is why the market is moving toward Word embedded copilots, playbooks, and review tools instead of pure self serve legal automation.
-
Spellbook learned this directly. Its earlier contract automation product tried to remove lawyers entirely, but users hesitated because they could not judge whether the document was actually safe. The current product keeps the lawyer in Microsoft Word, where they review redlines, generate clauses, and apply company playbooks inside an existing workflow.
-
The same pattern shows up across the category. Harvey is seeing strongest daily usage in transactional drafting and review, where repetitive work can be sped up by as much as 70%, while litigation remains harder because hallucinations and subjective judgment matter more. Legal AI is automating the first pass, not the final sign off.
-
Competition is splitting by how much human service stays in the loop. Spellbook sells software that helps in house and firm lawyers review contracts faster. Crosby goes further by delivering legal services with lawyer oversight. Luminance pushes furthest toward autonomous negotiation, but even there the product is aimed at high volume document flows inside enterprise legal workflows, not replacing counsel altogether.
From here, the category expands by moving one step upstream and downstream from redlining. The next layer is AI that triages intake, prepares a first draft, benchmarks terms against market data, and routes exceptions to a legal team. As models improve, more work shifts to software, but the durable products will be the ones that turn lawyers into managers of AI driven contract operations.