Connect Your AI Agent to Your Partner Program

What Is an MCP Server?

MCP — the Model Context Protocol — is an open standard for connecting AI assistants to the systems where your data actually lives. Instead of copying a CSV into a chat window and hoping the model reads it correctly, the assistant queries the source directly and acts on what it finds.

For a PRM, that distinction matters more than it might sound. Partner data goes stale fast: a deal moves stage, a partner is onboarded, a lead is registered. An export is a photograph of a moment that has already passed. An MCP connection reads the live record.

CinnaLab.io is one of the first PRM platforms to ship a native MCP server. Because MCP is an open standard rather than a vendor integration, nothing here locks you to a single AI provider — the assistant your team already uses is the assistant that connects.

What Your AI Can Do

Once connected, your AI agent can read across your partner program and take a defined set of actions inside it.

Read — List and inspect partners. View leads and the deal pipeline. Read your partner program setup, including tiers and commission structures. Pull program KPIs.

Write — Create and update leads. Register deals and update their stages. Invite new partners.

In practice that turns reporting questions into conversation. Which partners registered deals this quarter. Which leads from a given partner have not moved in thirty days. What our partner-sourced pipeline looks like against last quarter. Add a lead for a named account, sourced by a named partner. Register this deal and move it to Proposal.

These are questions a partner manager already asks every week. Today answering them means tabs, filters and an export somebody runs at month end. Connected, they are answered in the flow of the work.

What It Deliberately Cannot Do

Your AI cannot reconfigure your partner program. Tiers, commission rates and program rules stay human-only, by design and not by oversight.

The reasoning is straightforward. Reading a commission rate is useful. Changing one is a commercial decision with contractual consequences for every partner it touches, and it is not the kind of thing that should happen because an instruction was ambiguously worded. An assistant that can quietly restructure your commission model is not a feature.

This boundary is enforced at the server, not in a prompt. There is no combination of instructions that moves it.

Security and Tenant Isolation

Every connection runs on a token that a workspace admin generates in Settings. The token is displayed once and never shown again.

Scopes — Tokens are read-only by default. Write access is opt-in, granted deliberately rather than assumed.

Expiry — Choose 30, 90 or 365 days. Tokens expire on their own rather than living forever in a config file somebody forgot about.

Revocation — One click, immediate.

Audit trail — Every call your AI makes is logged, so the question of what an agent did is answerable rather than debatable.

Rate limiting — Applied per token.

Tenant isolation — Enforced on every request, and verified by a dedicated cross-tenant test suite that runs in continuous integration on every build. It is checked automatically rather than asserted in a sales conversation.

Setting It Up

Setup takes about two minutes and needs no developer.

1. In CinnaLab.io, open Settings → AI Connectors. 2. Generate a token. Choose its scopes and expiry. Copy it — this is the only time it is shown. 3. In your AI client, add CinnaLab as a connector and paste the token. In Claude this is a custom connector; in Cursor it is an entry in your mcpServers configuration; other MCP-compatible clients follow their own equivalent. 4. Ask your first question.

The AI Connector is available on every paid plan, including paid-plan trials. Free-plan workspaces can see it in Settings and upgrade to activate it. Token generation is restricted to workspace admins.

Who Gets The Most From It

Partner managers running a program day to day, who spend more time assembling the answer than acting on it.

Founders and revenue leaders who want a straight answer about partner-sourced pipeline without booking it as a reporting request.

Teams already building with AI agents, who would rather connect their PRM to the assistant they have than wait for a bespoke integration.

If your partner program currently runs on a spreadsheet alongside a CRM, this is the point where the reconciliation work stops being worth doing.