What was delivered
- An excel-to-powerbi skill for PI console: one line with the workbook and the report you want, in plain words.
- Workbook reading across multiple sheets: column names, data types and sample rows, sheet by sheet.
- Automatic relationships: the model finds the columns that match across sheets and joins them into a star schema.
- Measures written for you, such as revenue, order count, average order value and margin.
- A Power BI project (.pbip) built through the Power BI MCP server: the semantic model and the report pages, ready to open in Power BI Desktop.
How it works
- In PI console, type the skill name, the Excel file and what the dashboard should show.
- The local LLM opens the workbook and reads every sheet.
- It matches key columns across sheets and builds the relationships.
- It writes the measures and lays out the report pages.
- The Power BI MCP server writes the .pbip project, and the dashboard is ready to open.
In plain words
Turning a spreadsheet into a proper Power BI report usually means an analyst spending an afternoon on joins, measures and layout. Here you describe the report you want, and the local AI does that afternoon's work: it figures out how the sheets fit together and builds the dashboard for you.
Why it matters
Anyone who can describe a report can now get one, without learning data modelling first. The data never leaves the building, because the model that reads the workbook runs on the private local LLM. And the result is a normal Power BI project, so analysts can open it and refine it like any other report.