Connect enterprise localization to AI assistants and agents

Localization teams are spending more of their day inside AI tools. Engineers live in Cursor. Strategy teams use Claude. Product managers run half their workflow through assistants that can pull from Jira, GitHub, and Slack.

But not from XTM Cloud. Not until now.  

Today we're launching XTM MCP Server, a new way to connect AI assistants and developer tools directly to XTM Cloud.

If your team uses Cursor, Claude Desktop, VS Code, or any other MCP-compatible host, you can now query project data, check workflow status, and assign linguists, all in natural language, without leaving the tool you're already in.

It's available now to all XTM customers with API access, at no additional cost.

What is MCP, and why does it matter? 

Model Context Protocol (MCP) is an open standard that lets AI assistants connect to the tools and data your team already uses. Think of it as a standard way for AI assistants to securely connect with external business systems and data sources.

For localization teams, that's a meaningful shift. Until now, getting project information into an AI workflow meant copying it manually, building a custom integration, or relying on someone to summarize it for you. With MCP, your AI assistant can query XTM Cloud directly.

What you can do with

MCP today

The first release covers the most common questions teams ask about active work. From a connected MCP host, you can:

• Ask what's in flight: project status, workflow stage, assignments, due dates
• Pull linguist and reviewer information, including who's qualified for which language pairs
• Assign or unassign linguists to specific workflow steps
• Verify connectivity and basic system info to confirm everything's wired up

That covers a lot of the daily back-and-forth that usually pulls a project manager out of focus mode and into the XTM UI to look something up.

Who is MCP Server for?

MCP Server is built for the technically capable end of the customer base. If your team already uses the XTM API, this slots in alongside it. If your developers, automation teams, or innovation groups have been looking for ways to connect XTM into AI-driven workflows, MCP Server provides a structured way to do that.

A few example use cases we're already seeing:

• A localization manager in Claude Desktop asks, "What projects do I have due this week, and who's assigned?" and gets a structured answer pulled live from XTM Cloud
• A developer in Cursor building a localization automation script asks the assistant to verify which linguists are qualified for German technical content, then assigns them to the next workflow step
• An innovation team experimenting with AI-driven workflows uses MCP to give internal assistants access to live localization data, without relying on manual exports or custom-built integrations.

If your AI workflow needs live localization data, MCP gives it a clean way in.

How to get started 

Three things you need:

1. An XTM Cloud account with API access enabled
2. An API token (generated the same way you generate one for any other API integration)
3. An MCP-compatible host. Cursor, Claude Desktop, and VS Code are the most common starting points

From there, you point the host at the XTM Cloud MCP Server endpoint , paste in your API token, and start asking questions. Setup typically takes a few minutes.

What's coming next

This release is intentionally scoped. We've focused on the queries and actions teams ask for most, with rate limiting set at 10 requests per second per session to keep things predictable.

A few things on the near roadmap:

• OAuth 2.1 support, which will open MCP up to a much wider range of hosts beyond the current API-token-based setup
• Expanded endpoint coverage, so AI tools can interact with more of the workflow surface
• Deeper agent support, building on the foundation we've already laid with XTM Agent inside the platform

Together, MCP Server and XTM Agent give teams two complementary AI surfaces. XTM Agent brings agentic AI into XTM for in-platform tasks. MCP Server makes XTM available to AI agents living outside it. Most teams will end up using both.

The bigger picture

XTM is evolving toward more composable, AI-connected workflows that fit naturally into the tools teams already use every day, not just inside the platform itself.

MCP Server is a meaningful piece of that. It moves localization from a destination to a service your other systems can call. As more enterprise tools adopt MCP, that matters more.

For now, it's available, it's free with API access, and it works. Try it on a project this week and let us know what you'd like to see next.