As we move into advanced AI capabilities, A question has come up in relation to an MCP integration for AI. Is this on the roadmap for DataMiner, When might we see the first test cases?
Hey Ian,
We're definitely hard at work bringing LLMs to DataMiner! Since LLMs can only generate text, we need to connect them to "tools" so they can actually perform actions. Concretely, we're making sure they can:
- Fetch alarm data (Tool 1)
- Write and execute Python code to process alarms (Tool 2)
- Display trend graphs (Tool 3)
- …and more.
MCP is one way to give LLMs access to tools. However, it's not the framework we're currently using, as it requires more setup and effort. So, it's not on the roadmap for now.
Why might we switch to MCP later? Imagine we've released the DataMiner LLM Assistant and users want to connect their own tools to it. MCP would make that super easy. That’s why our architecture is already designed to allow a smooth transition to MCP if we see demand for it.
Finally, regarding test cases: some use cases were already showcased at IBC (join the Analytics forum next week if you want to see a demo :)).
So, when will you see those features appearing? Well, we're hard at work on an initial version of the DataMiner assistant which will analyze alarms, display trend graphs, parse pdf documents so users don't have to manually copy information from files into DataMiner,... A big part of that will be finished up this year with the goal of gathering early user feedback and discovering where to improve! For an official release, I'm aiming more towards February 2026!