Performativ connects bank staff’s AI assistants to client data under existing permissions
The MCP release is available to existing clients. Access is off by default, and the company says every request is logged.
Loading page…
The MCP release is available to existing clients. Access is off by default, and the company says every request is logged.
Listen to this story
Performativ made its next-generation Model Context Protocol (MCP) connection available to client firms on November 19, 2026, letting bank employees query client and portfolio records through assistants such as ChatGPT and Claude. Firms must enable the service and grant access; each employee’s existing Performativ permissions apply, and requests are logged. The release is for staff, not bank clients: CEO Albert Geisler Fox described client-facing portfolio queries as a future ambition.
The company describes workflows for client-review briefings, portfolio rebalancing discussions, and risk or compliance checks using records beyond holdings alone.
Performativ says its platform supports more than €100 billion in client assets, over 1,500 advisors, and connections to more than 150 custodians; these are platform figures, not MCP adoption numbers.
The MCP capability builds on governed agent workflows introduced with Performativ’s Custom Agents in November 2025.
Performativ wants banks’ clients to ask their preferred AI assistant about their portfolios. Its November 19, 2026 release starts with bank staff instead: the next generation of Performativ MCP connects assistants including ChatGPT and Claude to client records. The company says existing permissions and request logs govern that access.
The service became available to Performativ clients that day. It uses the Model Context Protocol, an open standard for connecting AI assistants to business systems. Rather than limiting questions to general investment information, the connection lets authorized employees ask about the client and portfolio records their firm holds in Performativ. The company describes uses spanning back-office teams, relationship managers, portfolio managers and compliance staff.
Performativ says the service is off until a firm enables it. Administrators decide who can use it, and each employee connects with an individual login and their existing Performativ permissions. That makes the employee’s authorized access—not an unrestricted view of the firm’s records—the stated basis for an assistant’s requests. Each request is logged for audit purposes, and Performativ hosts the MCP service in the EU.
Those controls address the problem Performativ sets out in its announcement: employees who manually copy client information into AI assistants move it outside the firm’s control. Its alternative is a connection to the underlying platform with the firm’s access rules applied. These are Performativ’s descriptions of the safeguards, including its claims about security and regulatory compliance.
The breadth of the connection comes from the records already held in Performativ. Firms use the platform for portfolio management, customer relationship management, compliance and reporting. According to the company, assistants can draw on holdings, performance and risk alongside projections, past advice, recent meeting notes, account activity and the identities of the advisors responsible for a client.
Performativ illustrates that scope with three staff workflows. Each combines portfolio information with the surrounding client record, rather than treating an investment account as an isolated set of positions:
The release builds on MCP capabilities introduced in November 2025 with Custom Agents, which lets firms design governed agent workflows inside Performativ. The new version brings that connection into the AI assistants employees already use. It changes where staff can ask questions, while drawing on the platform where their firm manages the client relationship.
Performativ says its wider platform supports more than €100 billion in client assets and over 1,500 advisors, with more than 150 custodians connected. Those figures describe the existing platform’s scale, not adoption of the newly released MCP service.
CEO Albert Geisler Fox described a further ambition in the launch announcement: letting banks’ own clients question their chosen assistant about a portfolio, with answers supplied under the bank’s controls. He framed that as a future direction. The access announced on November 19 is for staff at Performativ’s client firms.
Loading discussion...
Join the conversation
Explain which safeguard would matter most to your trust.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.