AI and MCP

I think there is a case for building AI into the TARGIT Windows and Web clients, allowing end users to ask natural language questions to one or more semantic models.

The output could be a TARGIT dashboard and/or report in one layout plus a natural language explanation of how this result was achieved and what the data actually says in another layout ... or something like that.

This is currently achievable by connecting e.g. Claude to an MCP server that's connected to a company's semantic models, but it would be a strong value-add if all of this could be done in TARGIT. Data access would of course be controlled by TARGIT's strong security model (this is an important differentiation from using Claude and MCP directly).

Context for the AI engine could be provided in Decorations or Language in TARGIT Management. By context I mean natural language explanations of dimensions, dimension attributes and measures that allow the AI engine to understand data and data relations in a semantic model ... this is to reduce hallucination.