Harvey has launched Harvey II, a new generation of its legal AI platform designed to embed artificial intelligence more directly into lawyers' daily workflows. The release introduces personalised memory, matter-based workspaces and more advanced legal agents, aiming to reduce the repetitive context-setting that has slowed AI adoption in legal work.
The update arrives as law firms push AI tools beyond standalone drafting assistance toward systems that understand individual matters and preferences. Harvey said the platform was developed through six months of testing with law firms and in-house legal teams across the US, UK, Europe, the Middle East and Asia-Pacific, with a focus on confidentiality, permissions and ethical walls.
Memory and personalisation
Harvey II's Memory function learns individual preferences such as drafting style, tone, structure and citation standards, then applies them across Harvey, Microsoft Word and Outlook. Users retain control over the information stored and can review, modify or disable the feature. Harvey said Memory data will not be used to train its global language models.
The rollout of Memory will take place in three stages, beginning with individual preferences, followed by matter- and team-level capabilities and, ultimately, organisational-level memory for law firms and legal departments. That staged approach gives legal teams time to assess how the system handles sensitive client information before wider deployment.
For legal professionals working across multiple practice areas, the system's ability to recall jurisdiction-specific citation rules and drafting conventions could reduce time spent re-entering the same instructions. Those working in document-heavy roles may find the integration with Word and Outlook particularly useful, since it moves AI assistance into tools already used for daily work. AI Learning Path for Paralegals covers similar workflow applications for support staff.
Spaces and legal agents
The platform also introduces Spaces, which organise documents, tasks, permissions and activity around specific matters. Legal agents can then use this context to perform more complex tasks and follow predefined workflows, rather than requiring lawyers to re-explain the background of each matter.
This structure mirrors how legal work actually operates - by matter, not by document. Organising AI context around individual cases or transactions means the system can track deadlines, relevant precedents and team permissions within a single workspace. AI for Legal resources cover how such matter-based tools fit into broader legal technology adoption.
Why this matters for legal professionals
The practical shift here is that Harvey II moves from treating AI as a tool you prompt to treating it as a colleague that remembers. For lawyers, that means less time repeating context and more time reviewing output - but it also means firms need clear policies on what the AI is allowed to remember, who can access that memory, and how ethical walls are maintained when agents work across multiple matters.
The phased rollout gives firms room to test those controls before organisation-wide deployment. Legal teams should use that window to review their own data governance rules and decide which matters are appropriate for AI-assisted workflows with persistent memory.
Your membership also unlocks:
