This article applies to client-side IT administrators and technical sponsors responsible for enabling PI's MCP connector during the current pre-beta phase.
Before enabling the Predictive Index MCP connector, please keep in mind the following prerequisites.
Note: The Predictive Index MCP is currently in pre-beta. We plan to make regular changes and updates on the road to its full release.
What "installation" means for a connector
Obi is enabled by adding PI’s remote MCP connector URL to a supported host and authorizing it with a PI login. There are two paths:
Individual / self-service — A single user adds the connector in their own host settings.
Admin-managed / org-wide — An admin publishes the connector to the whole organization or a group, and each member signs in individually.
Microsoft Copilot supports admin-managed rollout only. Claude and ChatGPT support both methods.
Who can enable PI's MCP connector
Individual: any user whose host plan permits adding custom connectors
Org-wide or group: a workspace or organization administrator of the AI host
Prerequisites for enabling PI's MCP
PI user provisioning. Every user must already be provisioned in PI with valid credentials and have Obi access enabled.
To create users in bulk, upload an organization file. This is our recommended flow, as it ensures accurate org data and makes Obi (and by extension the MCP) more effective.
To create users one by one, see our support article on creating and managing individual PI user records.
A supported AI host (Microsoft Copilot, ChatGPT, Claude) with permission to add custom or remote connectors. For admin-managed rollout, you'll need administrator access to that host’s workspace or organization settings.
Network access to PI’s connector and login endpoints — see our troubleshooting help article.
Additional support