Welcome to this week's blog spotlight: Turning FME Workspaces into MCP Tools: Connecting ArcGIS to AI. If you've been wondering how your existing FME workspaces could plug into AI agents without a rebuild, this one lays out the whole path.
Why This Matters
The blog reframes MCP as more than an AI feature: it's a general-purpose integration layer. Connect a data source or workflow once, and every MCP-capable client, whether that's a chat assistant, an app, or another system entirely, can use it. For FME users, that means years of accumulated workflows don't need to be rebuilt to become AI-ready; they just need to be exposed.
Inside the Blog
The post walks through turning a municipal sidewalk-repair workflow into a live AI tool, then scales up to multi-tool, agent-driven workflows. A few highlights:
- Any FME workspace can become an MCP tool in three steps: swap the terminal step for an MCP writer, publish to FME Flow, and register it with clear parameter descriptions.
- The tool's description is effectively its API contract. Since the agent never sees the underlying workspace, it needs explicit valid values, formats, and instructions for when to ask the user instead of guessing.
- Small, single-purpose tools beat monolithic ones. An agent can chain focused tools in combinations you never explicitly designed.
- Pushed further, a "planner-in-a-loop" workspace lets an AI accumulate facts and answer open-ended questions across changing datasets, without any logic being rebuilt for each new question.
Key Takeaways
The governance angle stands out. FME still handles authentication, permissions, and logging on every MCP call, and the AI never touches raw data directly, only the results a tool chooses to return. That "data guardian" framing means opening workflows up to AI agents doesn't mean giving up control over what's exposed.
Join the Conversation
If you converted one of your own workspaces into an MCP tool, what would it do? Is there a workflow sitting in your library right now that's just begging to become an AI-callable tool? Let us know below!
See you next time for another blog spotlight!