Using MCP with AI

MCP or the Model Context Protocol is an open standard that gives AI model a universal way to connect to external data sources including the PIMInto PIM and read from and write to the PIM depending on the permissions provided.

This means that AI can share data between multiple systems that support the Model Content Protocol making it very easy to extract data from one system to use for enrichment or updating information in another system.

The MCP Server is located at https://{myinstance}.piminto.com/mcp

The server requires that an APIKKEY is included in the HTTPS Request Header and a paid premium plan or greater.

How to add an MCP server to ChatGPT

1. Open ChatGPT in your browser.

2. Click your profile icon and select Settings.

3. Go to Apps & Connectors (or Plugins, depending on your interface).

4. Look for Advanced settings or Developer mode.

5. If available, enable Developer mode.

6. Select Create app or Add custom connector.

7. Enter your MCP server's URL and configure authentication if required.

8. Save the configuration and test the connection.

The exact options depend on your ChatGPT plan, workspace permissions, and available features.

Add MCP to Claude

In Claude, an MCP server is added as a custom connector. You need the server’s remote URL to set one up.

Pro or Max plans:

1. Go to Settings → Connectors.

2. Click “+” then “Add custom connector.” (see caude help here)

3. Add your connector’s remote MCP server URL.

4. Finish configuring your connector by clicking “Add.”

Usually you’ll then go through an authentication configuraton process to add an APIKEY to the MPC Header configuration.  This will securely sign in to your instance of the the application and grant specific permissions. After that, switch it on in a chat from the tools menu.