Crosby's Workflow Moat for Contracts
Crosby
Crosby will keep winning only if it owns the workflow around startup contracts, not the underlying model. The same legal reasoning is rapidly becoming available to every legal AI company, which shifts competition toward who is embedded in the actual job, intake, review, redlines, approval, and delivery of legal advice. Crosby is better insulated than pure software tools because it sells completed legal work with lawyer oversight, but that moat is operational, not model based.
-
Harvey is the clearest proof point. After frontier models caught up on legal reasoning, Harvey moved away from a proprietary fine tuned model and toward agentic workflows, customer success, and distribution, while still scaling to an estimated $195M ARR in 2025. That shows the value is moving above the model layer.
-
In legal AI, products are fragmenting by concrete workflow. Crosby handles startup contract review as a law firm. Spellbook sells drafting and review software to lawyers. Ironclad owns the contract system of record and approval workflow inside companies. When models commoditize, the company closest to the daily workflow usually keeps pricing power.
-
Ironclad shows why workflow can outlast model novelty. Its stickiness comes from being wired into approvals, versioning, renewals, and cross functional contract processes, not from having the smartest model. Clio is making a similar move by combining practice management with legal research through vLex. The durable moat is owned distribution plus proprietary workflow data.
The next phase of legal AI will reward companies that turn cheap reasoning into faster, safer, repeatable legal work. For Crosby, that means building the default path for startup contracts, with intake forms, playbooks, jurisdiction coverage, and lawyer review loops that get better with every matter. If that layer hardens, commoditized models become a cost advantage, not a threat.