Article Summary
AutoRFP has two separate MCP (Model Context Protocol) integrations that sound alike but work in opposite directions: MCP Connectors let AutoRFP's own agent pull data in from other tools, and the AutoRFP MCP Server lets outside AI assistants pull data out of AutoRFP.
Estimated Time
1 minute read
The Two Directions
Term | What it does |
MCP Connectors | AutoRFP's own agent reaches out to another service, such as GitHub, Grain, HubSpot, Intercom, Linear, Notion, Slack, or a custom server, so it can use that service's data while answering or drafting. Set up under Organization → Integrations → Agent MCP Servers. |
AutoRFP MCP Server | An outside AI assistant, such as Claude, ChatGPT, or Microsoft Copilot, reaches in to AutoRFP to read your projects, requirements, tags, and content library (read-only). This is set up on the assistant's side (e.g. in Copilot Studio), not inside AutoRFP. |
Where to Find Each One
Setting up a connector so AutoRFP can use another tool's data: see How to Set Up MCP Connectors in AutoRFP.
Letting an AI assistant pull from AutoRFP: see How to Connect to AI Assistants (MCP Server), or for Copilot specifically, How to Integrate with Microsoft Copilot.
✋🏼 Common Mistake to Avoid
Looking for "Copilot," "Claude," or "ChatGPT" inside AutoRFP's Tools / Agent MCP Servers list — it won't be there. That list is only for connectors AutoRFP pulls data from. Connecting an AI assistant to AutoRFP is configured on the assistant's own side instead.
Need Help?
💬 Live Chat: Available in-app
📧 Email: [email protected] or contact your Success Manager directly for urgent support.
📚 Learning Centre: learn.autorfp.ai/en