Harvey has made its integration with Intapp Walls for AI generally available after an early-access period. The connector allows law firms to bring existing ethical-wall policies into Harvey and enforce them across the legal AI platform without maintaining a second set of access rules.
The release addresses a specific governance problem for legal AI adoption. Law firms already use conflict systems and document controls to restrict access to sensitive matters, but an AI workspace can create a separate route to the same information if those policies are not applied consistently. Harvey and Intapp say their integration brings that enforcement into Harvey’s own collaboration and analysis surfaces.
How the Intapp integration works
Firms can synchronise policies from Intapp Walls for AI into Harvey. The controls apply across Harvey surfaces including Threads, Vault, Review Tables and Shared Spaces. Harvey says policies also govern collaboration features such as Workflows and Playbooks, so the same matter restrictions follow content as lawyers work across different parts of the platform.
The connector is designed to inherit rules from Intapp rather than ask administrators to recreate them in Harvey. When an authorised-access decision cannot be confirmed, Harvey says the system blocks access. This fail-closed approach is important for firms that need controls to remain effective when synchronisation or identity data is incomplete.
Policy changes in Intapp are synchronised automatically. In practical terms, a lawyer should see only the Harvey resources permitted by the firm’s existing wall configuration. That can reduce manual administration and lower the risk that a separate AI configuration drifts away from the source conflict system.
Audit records across both systems
The generally available integration also creates records intended to support compliance reviews. Intapp logs wall configuration and policy sources, while Harvey monitors synchronisation cycles, access attempts and enforcement actions. Together, the vendors say these records provide an evidence trail for internal governance, regulatory review and legal proceedings.
An audit trail does not by itself prove that every policy is correctly designed or every identity is accurate. Firms will still need processes for testing wall creation, user changes, emergency access, failed synchronisation and the retention of logs. They should also confirm how records are exported and how long each vendor retains them.
The announcement says Intapp supplies ethical-wall infrastructure to more than 70 per cent of the AmLaw 100. That vendor-reported adoption figure helps explain the integration’s relevance for large firms, but it does not remove the need for each organisation to validate its own configuration and professional obligations.
Why ethical walls matter for AI agents
The access-control problem becomes more important as legal AI moves from isolated prompts towards agents that can retrieve documents, analyse matters and execute multi-step work. A user may be authorised to operate Harvey generally but prohibited from seeing a particular client matter. The agent must respect the same boundary when searching, summarising, sharing or generating work based on connected content.
Central policy inheritance can make those restrictions more consistent. It can also help collaboration: authorised lawyers can use Shared Spaces and other Harvey tools without administrators rebuilding permissions for every project. Sensitive matters should remain inaccessible to screened users while permitted teams continue to work across practice groups.
Firms should nevertheless assess controls beyond ethical walls. The announcement does not replace broader decisions about data residency, model providers, retention, privilege, human review or which tasks agents may execute. Organisations also need to consider information brought into Harvey from systems not governed by Intapp and content users create directly within the platform.
Licensing and deployment considerations
The integration is generally available, but it is not included automatically for every Harvey or Intapp customer. Harvey says firms need an Intapp cloud licence and a separate Harvey connector licence. Interested customers are directed to their Intapp representative for activation details.
Before enabling the connector, administrators should map users and matters between systems, identify the source of truth for identities, and test both allowed and denied scenarios. A staged rollout can help confirm that policy updates arrive promptly and that fail-closed behaviour does not interrupt authorised work unexpectedly.
Security and compliance teams should also define monitoring responsibilities. If synchronisation fails, they need to know which system raises the alert, who investigates it and what evidence is available. Lawyers and support staff will need guidance on how to report access that appears too broad or too restrictive.
What general availability changes
The move from early access to general availability gives law firms a supported route to extend a widely used conflict-control system into Harvey. It is a meaningful enterprise capability because it connects legal AI governance to policies firms already operate, instead of relying solely on separate workspace settings or user discipline.
The practical outcome will depend on implementation quality. Firms should verify synchronisation, audit coverage, licensing, support arrangements and controls for all connected data before expanding agent use. For Harvey, the release strengthens its governance story as legal AI systems take on more work across documents, teams and matters.