Articles in this section

Connect Factbird to external AI tools with a Model Context Protocol (MCP) Server Connector

Factbird can connect to external AI harnesses like Claude, ChatGPT and Copilot.

What is an AI harness?

Large Language Models by themselves only generate text. They need a harness to gain access to tools to do things like search the internet, use connectors like MCPs and operate in the real world.

Read more: What is an AI harness?​

Our MCP in action inside Claude Desktop. Here we have asked the MCP for our top 5 stop cause reasons in the past month. The agent is smart enough to know to use the Factbird MCP without being asked. The Claude Desktop harness supports the MCP-UI standard, and can generate Factbird native Customizable Reports charts.

What the MCP unlocks

MCPs enable you to:

  • query Factbird data with external AI harnesses like Claude, ChatGPT and Microsoft Copilot.
  • query Factbird data alongside other connected data sources like ERP systems easily.

For example: "Am I going to hit my promised delivery timelines this month" may require querying ERP system’s MCP for the deliveries promised across multiple production orders to compare with Factbird production output data.

  • chat back-and-forth about results. You can ask follow-up questions like “Investigate why we may be experiencing an increase in stop type [X]” or "slice this by product produced"
  • specify an output format for your analysis flexibly. You can ask to "Export this analysis to Excel" or "Send this in a PowerPoint presentation to my manager" provided that you have the right tooling to do these tasks.
  • in some harnesses, like Claude Cowork and ChatGPT, you can set up scheduled tasks for repeated analysis.
  • stay at the frontier of AI tooling development. New developments in AI tools take time to find their way to our on-platform assistant.

Compatibility

The following harnesses are currently compatible with Factbird’s MCP implementation:

  • Claude (web, Desktop and mobile). Set up the connector on claude.ai.
  • ChatGPT (web, Desktop and mobile for workspace accounts). Personal Plus and Pro accounts use developer mode, which works on the web only.
  • Microsoft Copilot (Teams, Microsoft 365 Copilot and Copilot Studio)
  • Other AI harnesses that support MCP OAuth (CIMD or DCR) such as Google Antigravity 2.0

Installation

Installation instructions vary depending on the account role that you hold with your AI provider. You will need to provide the following URL where connections are managed in your AI harness.

https://api.cloud.factbird.com/mcp

Adding a connector to an external AI harness like Claude, ChatGPT or Microsoft Copilot requires administrator access for a corporate account’s workspace or a personal account with the correct plan for that tool:

  • Claude: Free (one custom connector only), Pro, Max, Team or Enterprise.
  • ChatGPT: Plus, Pro, Business, Enterprise or Edu.
  • Microsoft Copilot: a work Microsoft 365 Copilot licence with Copilot Studio access.

When the connector is added, users of the connector log in with their existing Factbird account:

An example showing MCP authentication triggered via Claude on the Factbird platform. A logged in user must click ‘allow’ to use the MCP with their harness of choice.

Business subscriptions (install for the whole organisation)

Install the Factbird MCP in Claude (business subscriptions)

Install the Factbird MCP in ChatGPT (business subscriptions)

Install the Factbird MCP in Microsoft Copilot (business subscriptions)

Personal subscriptions (install for yourself)

Install the Factbird MCP in Claude (personal subscriptions)

Install the Factbird MCP in ChatGPT (personal subscriptions)

Install the Factbird MCP in Microsoft Copilot (personal subscriptions)

Other AI harnesses

The Factbird MCP works with any AI harness that supports remote MCP servers with OAuth sign-in through Client ID Metadata Documents (CIMD) or Dynamic Client Registration (DCR). Add the MCP URL where your harness manages connectors, then sign in with your Factbird account. Check your harness documentation for the exact steps.

Advanced MCP usage

Automate analysis and get daily digests with scheduled tasks:

Advanced MCP usage: Automate analysis, send and receive daily digests via Claude Cowork

Advanced MCP usage: Automate analysis, send and receive daily digests via ChatGPT

See also:

What is an AI Harness?

Supported Queries and Usage Advice for the Factbird MCP

Steps Taken to Reduce the Likelihood of AI-Generated Mistakes When Using the MCP

Factbird MCP: Frequently Asked Questions

Was this article helpful?
0 out of 0 found this helpful