Harvey

Harvey is a legal AI platform for law firms and in-house teams, and this launch introduces Harvey II, the second generation of the product. Harvey II builds on Harvey's earlier announcement that it was building Memory , adding matter-based workspaces and a new in-house model called Harvey Tenet. The platform combines memory of how lawyers work, matter spaces, and smarter agents so that when an agent picks up a task, it already has the files, permissions, and prior context tied to that matter or project. Lawyers can assign work to human colleagues or to agents from the same interface and track it end to end. The centerpiece is Harvey Tenet, the company's first model, post-trained end-to-end for legal reasoning , which Harvey positions as frontier-level on prominent legal benchmarks and priced closer to open-source models, making it practical to run agents continuously across every matter . Memory captures individual preferences like structure, citation style, and preferred phrasings, while firm-level templates such as heavily negotiated credit agreements or approved fallbacks let drafts come back in the shape a team already uses. Harvey says the same setup lays the groundwork for partnering with organizations to build specialized models around their own legal work , so two firms on the platform will end up with different behavior over time. Harvey was founded in 2022 by Winston Weinberg and Gabriel Pereyra, who started the company as roommates in a San Francisco apartment, with Pereyra coming from AI research at Meta and Google and Weinberg from legal practice . This release matters now because the legal AI market is shifting from single-shot assistants toward agents that run in the background on real matters, and Harvey II is the company's attempt to own that shift with a purpose-built model and a workspace model organized around how lawyers actually structure their work.