Anthropic Competes Directly with Harvey

Anthropic moving into legal means the model layer is no longer just Harvey’s supplier, it is becoming a substitute. Harvey still sells a law firm ready product with legal research, document review, and enterprise controls, but Anthropic can bundle legal work directly into Claude, the same way model companies moved from powering coding startups to launching their own coding products. That makes Harvey’s edge less about raw model quality and more about workflow, security, and trusted integrations.1 2 4 5

• Harvey’s product is built for lawyers doing research and document review, and it has scaled fast, from an estimated $50M ARR at the end of 2024 to $400M by August 2026. That growth shows real demand, but it also makes the company a large target for platform vendors that want to own more of the application layer.4
• Anthropic is not testing one narrow feature. Its legal push is described as a suite of more than a dozen plug ins for contract review, NDA triage, deposition prep, compliance workflows, and legal briefings. That overlaps with the day to day tasks where Harvey wins seats inside firms and corporate legal teams.2 5
• This follows the same pattern seen in coding. Anthropic used Claude as the model behind AI coding products, then launched Claude Code and Claude Cowork itself. Once the foundation model company ships the interface, the vertical startup has to defend the workflow layer, not just access to the best model.5 6 7

The next phase of legal AI will split between broad model suites and specialist workflow products. Anthropic and other labs will keep absorbing generic drafting and research tasks, while Harvey’s path is to become the system firms trust for matter specific workflows, security, and integrations that are painful to rebuild in house.1 2 5