Write data back via MCP

Support for working with your training data through MCP (the Model Context Protocol) is coming soon. It will let AI assistants and agents read your Coassemble results and write them into the tools you already use, so reporting and record-keeping can happen through the AI tools you work in. This page describes the intended approach and will be updated with exact steps when it ships.

MCP is an open standard for connecting AI assistants and agents to external tools and data. Working with your results through MCP means an AI agent could read your Coassemble training data and write it into the systems where you keep records, rather than you exporting it or building a custom integration for each tool.

What this is for

Training results are most useful when they reach the systems your team actually works in: a CRM, an HR system, a data warehouse, or an internal dashboard. Today you get results out by exporting a CSV or pulling them through the API. MCP is intended to add a third way, where a connected AI agent can move the right data to the right place for you, and keep records updated, without a separate build for every destination.

How it's intended to work

The plan is for Coassemble to make your training data available over MCP so a connected AI assistant or agent can read results and write them into your other tools. Because MCP is an open standard, this is intended to work across the AI tools that support it rather than being tied to a single assistant.

The exact setup, what's required, and which plans include it will be confirmed when the feature is released. We'll update this page with concrete steps then.

Frequently asked questions

Is this available now?

Not yet. Working with your data through MCP is coming soon. For now, you can export results as a CSV or pull them through the API. This page will be updated with setup steps when it's released.

What is MCP?

MCP, the Model Context Protocol, is an open standard for connecting AI assistants and agents to external data and tools. It lets an AI tool read your results and write them elsewhere without a one-off custom integration.

Which tools will this work with?

MCP is an open standard, so the intent is to support the AI assistants and agents that work with it. Specifics will be confirmed at release.