Smarsh adds AI tools for governed communications data

Smarsh has expanded its Communications Intelligence Platform with a new MCP Server, AskSmarsh AI and additional software agents. The launch targets regulated industries that manage large volumes of communications data.

The update adds a Model Context Protocol server that connects existing artificial intelligence tools with Smarsh's archive and intelligence systems under existing governance rules. It also introduces a conversational interface, AskSmarsh AI, that lets users search, investigate, report and export communications data through a single prompt-based workflow.

The changes are intended to replace manual work spread across separate archives, point products and vendor-specific interfaces. Smarsh says compliance, legal and supervisory teams can work within familiar tools rather than move data into separate systems.

The MCP Server is central to that approach. Smarsh says it gives AI tools a policy-controlled way to query archived communications, surface insights and trigger workflows while maintaining the role-based access controls, audit logs and chain-of-custody requirements customers already use.

That matters for firms in banking, insurance and financial advice, where internal communications are often subject to retention, monitoring and disclosure rules. Efforts to use AI on that data have raised concerns about whether records remain governed and whether access controls can be preserved when information is moved into new systems.

Kamesh Tumsi, Chief Product Officer, said the expansion reflects a change in how customers want to use communications data.

"AI changes the model," said Kamesh Tumsi, Chief Product Officer, Smarsh.

"Customers will be able to use AskSmarsh AI to interact with their communications data conversationally, build on top of us, or bring governed Smarsh intelligence into the tools they already use. We're building for the challenges our customers face today while giving them the openness, intelligence and flexibility to adapt to what comes next," Tumsi said.

Agent tools

AskSmarsh AI is built as a conversational layer over the wider platform. It uses separate search, reporting and export agents behind the scenes to reduce the manual query building and system switching often required in investigations or reporting tasks.

Smarsh is also extending its broader set of agents. MySearch Agent is designed to give employees self-service access to their own archived messages across email, chat, voice, mobile and social channels without going through IT or compliance teams.

Discovery Agent adds legal research functions focused on financial misconduct, non-financial misconduct and employment discrimination. Smarsh says early case assessment now runs directly within the archive using what it describes as a Semantic Context Graph.

Another product, Intelligent Agent, is being updated with three new models for non-financial misconduct: hostile, abusive and discriminatory language; sexual harassment; and intent to resign. These additions broaden the platform's monitoring scope beyond market conduct and other traditional financial supervision use cases.

Wider integrations

Smarsh is also broadening the systems from which customers can capture data. New integrations cover Claude Enterprise, Amazon Bedrock and Google Workspace, extending the types of AI and workplace communications that can be brought into the platform's archive and oversight processes.

The move reflects a wider shift in regulated sectors, where firms are trying to apply generative AI and automation to compliance and legal processes without weakening recordkeeping controls. Across archiving, e-discovery and surveillance, vendors have increasingly focused on embedding AI into governed workflows rather than treating compliance archives as static storage systems.

Smarsh has long sold communications capture, retention and oversight products to regulated organisations. Its client base includes nine of the top 10 banks in North America, Europe and Asia, along with brokerage firms, insurers, registered investment advisers and government agencies in the United States.

The latest expansion suggests Smarsh wants to move beyond archiving and monitoring into a broader role as an intelligence layer for communications data. By linking archives to external AI tools and adding native conversational and agent-based interfaces, it is seeking to make compliance data more directly usable across legal, supervisory and employee workflows.

According to Smarsh, every interaction through the MCP Server remains subject to the same role-based permissions, audit logging and chain-of-custody controls already in place.