Legal AI startup Harvey has announced plans to roll out a new feature called Memory, positioning it as a way for law firms and corporate legal departments to retain institutional knowledge inside its platform. The company disclosed the initiative on Thursday in comments to Legaltech News.

Memory is intended to function as an opt-in layer within Harvey’s existing legal workflow tools. According to the company, it can retain drafting habits, stylistic preferences and matter-specific context so that future work reflects prior decisions without requiring lawyers to restate instructions or search for precedent.

In practice, that could include recalling a client’s preferred tone in transaction documents, resurfacing previously approved clauses or aligning new drafts with firm-wide best practices. The aim, Harvey says, is to reduce repetitive setup work while keeping legal judgment in human hands.

Harvey co-founder and president Gabe Pereyra framed the tool as a way to scale expertise that is often concentrated in a small number of senior lawyers. “The best partners at a law firm are constrained in how much they can train every associate,” he said, adding that a shared memory layer could make that expertise more accessible across teams.

The company says Memory is currently being developed in collaboration with customers, including both law firms and in-house legal teams. That co-building process is focused not only on functionality, but also on administrative controls and ethical guardrails—areas that remain sensitive as generative AI becomes more embedded in legal practice.

Those concerns mirror broader debates across the profession about confidentiality, data segregation and professional responsibility. Harvey has said it is working with its internal innovation team and external users to determine how memory systems should operate in a regulated environment, where the tolerance for error is low and accountability is personal.