A person in a light blue shirt typing on a laptop with digital data displayed on the screen.

INSIGHTS

How FCM fits into the AI tools you already use  

FCM's Model Context Protocol (MCP) connects FCM's travel and reporting intelligence directly to the AI assistants your teams already use, including Microsoft Copilot, Claude, and ChatGPT.  

Instead of logging into a separate platform, your travellers and travel managers can book trips, pull reports, and generate policy recommendations through a single conversational interface. 

 

Key Takeaways 

  • FCM's MCP server connects travel booking and reporting data directly to the AI tools your organisation already uses 
  • Travellers and arrangers can search, book, and manage trips in natural language, without leaving their preferred AI assistant 
  • Travel managers and analysts can query spend data, generate rate recommendations, and draft policy memos through a single conversational thread 
  • FCM's governed data definitions and Power BI semantic layer ensure the numbers you get are accurate and agreed upon, not derived on the fly 

Contents

AI has moved on, and so has FCM 

Most AI tools started as standalone assistants. You'd ask a question in a chat window and get an answer; useful but limited.  

The shift with MCP elevates those tools so they become context aware. It means they know your data, definitions and permissions before a question is even asked.  

FCM's MCP sits between its platform, and the AI tools your team already use – such as your company's version of Claude and Copilot. It means that search, booking, policy, data and analytics can sit in those tools, with FCM's intelligence underneath it.  

How FCM approaches MCP 

FCM's MCP servers connect to the same FCM intelligent layer, the same governed data, and the same security standards that are applied to FCM Platform. Every query is logged. Metrics are pre-defined.  

The result is an AI connection that's structured, auditable, and built for enterprise use, not just a proof of concept. It's available in any language, and defaults currency per user.  

Safety and duty of care functions are excluded, because AI is non-deterministic and traveller safety can't be. 

Use case one: Search and book 

A traveller is in their employers' tool, Copilot or Claude, mid-morning, preparing for a client trip later that year. Rather than opening a separate booking tool, they go through these steps still within that same tool:  

  • Ask: "Find me a return flight from London to Frankfurt on November 24th, returning Thursday 26th, within policy." 
  • Decide: Compliant options based on corporate policy, traveller’s profile and past travel booking behaviour are shown to the traveller, who can confirm either in the tool, or can click through into the more traditional online booking tool interface.  

No tab switching. No separate login. No copy-pasting itinerary details into a calendar invite. 

For travel arrangers managing bookings on behalf of others, the same logic applies. They can ask, confirm, and action, all from within the tools they're already using to manage their day. 

Use case two: reporting 

Travel managers have long relied on dashboards and pre-built reports. Those tools are valuable, but they're limited. FCM's MCP changes that.  

  • Ask: A travel manager working in Claude or Copilot can ask: "Show me our hotel spend by city for the last 12 months." The assistant returns the data, already defined using FCM's governed metrics, not raw tables that need interpretation. 
  • Decide: Continue the conversation: "Recommend a three-tier nightly rate cap based on our actual spend, with justification." The assistant analyses the data and returns a tiered recommendation with a written rationale, grounded in the organisation's own booking history. 
  • Act: "Draft a traveller memo and an executive summary for the CFO." Both produced, in the same thread. 

This is what FCM describes as moving from a relationship with data to a conversation with it. The insight arrives already interpreted. The decision, and the communication around it, can follow immediately. 

FCM's MCP integration puts travel intelligence where your teams already work. Fewer logins, faster answers, and data your organisation can act on without setting it up from scratch every time. 

To see how FCM MCP fits into your specific tech environment, book a demo with the FCM team. 

Frequently Asked Questions 

  • What is FCM's MCP and how does it work? 

    FCM's Model Context Protocol (MCP) is a server that connects FCM's travel and reporting capabilities to external AI assistants, including Microsoft Copilot, Claude, and ChatGPT. When a user asks their AI assistant a travel or data question, the assistant calls FCM's MCP server, retrieves the relevant information or performs the relevant action, and returns the result within the same conversational interface. 

  • Which AI tools are compatible with FCM's MCP integration? 

    FCM's MCP server is compatible with any AI assistant that supports the MCP standard, including Microsoft Copilot, Microsoft Teams, Claude, ChatGPT, and Perplexity. 

  • Is FCM's MCP integration secure? 

    Yes. FCM's MCP server meets FCM's enterprise security standards. Every query and response is logged for oversight. The server has been penetration tested and fully security reviewed. Duty of care functions are deliberately excluded from the MCP layer, since those require deterministic tooling. 

  • Do I need a separate licence to access FCM's reporting through MCP? 

    The travel (Sam) MCP server and the reporting MCP server are managed separately. The reporting MCP requires a Power BI licence; the travel MCP does not. Your FCM account team can confirm what applies to your organisation's setup. 

  • Can FCM's MCP connect to data sources outside of FCM? 

    Yes. FCM's MCP can be linked to other MCP-compatible endpoints, such as expense platforms, ERP systems, or HR data. This makes it possible to ask questions that span multiple data sources, for example, whether travel frequency correlates with sales outcomes, without building a separate data warehouse or running an integration project. 

  • How is this different from what we can already do with FCM's reporting dashboards? 

    FCM's dashboards provide governed reporting, benchmarking, and self-service views. They're strong for anticipated questions. The Reporting MCP extends that foundation to unanticipated questions, allowing users to query data conversationally, receive interpreted answers, and produce deliverables (such as memos or executive summaries) in the same thread. The two work together rather than in competition.