The MCP Connection, Explained

What is it?

The Model Context Protocol (MCP) is an open standard that lets an AI assistant such as Claude connect directly to a specific system — in this case, Corporify — and query it live, on request.

Once connected, Claude can look up governance, entity and document data stored in Corporify while you're chatting with it, instead of you exporting or copy-pasting it in manually. It is read-only and on-demand: Claude fetches only what a given question requires, in the moment, and does not hold a standing copy of your Corporify data. 

Who has access?

Activation is layered, not a single switch. First, the customer organization approves switching MCP on for their Corporify account. Separately, the organization's own admin for the connecting AI tool (e.g. a Claude admin) has to approve and configure the connector on their side too — that approval is independent of Corporify's. Only once both are in place, can individual users connect, each with their own Corporify credentials, and each still limited to what they're already authorized to see inside Corporify — the same entities and permission scope as when they log into the platform directly.

No new access is created by adding Claude or any other AI tool. It simply gives an already-authorized user a faster way to ask for information they could already retrieve manually. 

Setup steps

• The customer organization first approves activating the MCP connection on the Corporify side.
• Separately, the organization's admin for the connecting AI tool (e.g. a Claude admin) must also approve and configure the connector on their end before anyone can use it.

• Once both approvals are in place, each individual user can activate the connection in their own AI tool. The exact setup differs per AI tool — see the Corporify knowledge base for tool-specific instructions.
  

What does it mean for you?

Faster, conversational access to your own Corporify data — governance, structure, and document look-ups without leaving the chat or logging into the platform separately.

Nothing changes about where the data lives: Corporify remains the single source of truth. The AI tool is a read-only reporting layer on top of it — it cannot create, edit, approve, sign, or delete anything. Every change still goes through Corporify's normal workflows.

Security

• Read-only by design: only “get” and “search” operations are exposed (company info, governance, shareholders, UBOs, delegation of authority, investments, people, and documents). No create, edit, or delete capability is available to the AI tool.
• Scoped to the user: access always mirrors the connecting person's existing Corporify permissions; nothing beyond that is reachable.
• Layered approval: a company-level approval on the Corporify side is required, the connecting AI tool's own admin must separately approve and configure the connector, and each user then activates it individually.

Admin-configurable scope: in Claude specifically, the admin can narrow this further: e.g. disabling one data point (such as UBOs) while leaving the rest available, or restricting connector setup to admins only. Equivalent controls in other AI tools may differ. 

Use cases

• “Who are the UBOs of [entity]?” — an instant answer, without opening the portal.
• Prepping for a board meeting: a quick summary of an entity's governance structure and delegation of authority.
• Due diligence / KYC prep: pulling shareholder structure and investments across a group in one pass.
• Group structure overviews: quickly mapping entity relationships across a portfolio for internal reporting.