Power BI Modeling MCP Server: connecting AI to your local data model
Power BI Modeling MCP Server: why does it matter?
Power BI Modeling MCP Server is a new way to connect AI directly to your Power BI data model.
Any BI or data analyst knows that the more columns and rows you keep, the heavier the file becomes, which can hurt performance. But what if we could apply performance best practices automatically with the help of AI?
What is Power BI Modeling MCP Server?
In simple terms, Power BI Modeling MCP Server is a local server that bridges AI agents and your Power BI semantic model. Instead of having AI read only plain text or disconnected spreadsheets, it can inspect the model itself. That includes tables, relationships, measures, and how the business logic is structured.
It also uses the MCP specification, Model Context Protocol, to standardize communication between the AI and the model. That gives the agent much better context and allows it to suggest improvements, create measures, generate documentation, and even help with performance optimization.
Local connection without depending on Microsoft Fabric
An important point here is the local workflow. You run the MCP Server locally and connect it to Power BI models that are open on your computer. The AI talks to the server, and the server talks to the model.
How can it help?
Power BI Modeling MCP Server acts like a modeling copilot for BI analysts. It reviews the model, suggests best practices, identifies unnecessary columns, incorrect relationships, and tables that can be simplified. That makes semantic model development faster and more consistent. It can also create standard DAX measures while explaining what each one means in business language.
Another useful capability is automatic documentation. The AI can generate descriptions for tables and columns, summarize the purpose of the model, and keep documentation updated as structures and metrics evolve.
It can also automate recurring tasks, such as generating a calendar table or adjusting data types, through natural language commands. The agent applies the changes to the model, even though human review can still remain in the loop.
At the moment, the MCP Server works only on the semantic model itself, without support for creating charts or visuals.
How this changes day-to-day work
Today, the typical workflow is to open a .pbix file, wait for it to load, test an adjustment, save, reload, and repeat. With Power BI Modeling MCP Server running locally, that workflow becomes much smarter. You connect AI to the model, describe what you want to improve, and receive targeted suggestions.
You keep the final decision, but you gain a specialized partner looking at the structure of your model rather than only the raw data. That makes the process faster, reduces errors, and improves modeling decisions.
How to get started
Tools you need to get started:
- Download and install VS Code
- With VS Code open
- Install the extensions GitHub Copilot and GitHub Copilot Chat
- Install the extension Power BI Modeling MCP Server
- You are ready to begin
Keep your .pbix file open, then go to VS Code, click the Chat icon, and type the following command:
Connect to Power BI Desktop "your pbix file name" locallyWait a few seconds and confirm the action. If the result is positive, you can start using the available functions.
In my case, I use the paid GitHub Copilot plan and have access to several models. In the example, I was using Claude Sonnet 4.5.
Prompt and action ideas
Here is a brief list of things you can ask for.
| Possible scenario | Prompt |
|---|---|
| Review naming conventions and rename in bulk | Analyze the naming conventions in my model and suggest renames to guarantee consistency. |
| Add descriptions across the model for documentation | Add descriptions to all measures, columns, and tables to explain their purpose clearly and describe the DAX business logic in simple language. |
| Translate your semantic model | Generate a French translation for my model, including tables, columns, and measures. |
| Refactor queries to use semantic model parameters | Analyze the Power Query code for all tables, identify the data source configuration, and create semantic model parameters to make the source location easy to switch. |
| Benchmark DAX queries across multiple models | Connect to the semantic models "V1" and "V2" and benchmark the following DAX query on both models. [DAX Query] |
| Document your semantic model | Generate a Markdown (.md) document with full professional documentation for a Power BI semantic model. Use a simple mermaid diagram for table relationships, document each measure including DAX code and business logic, document row-level security, and document data sources by analyzing Power Query. |
Do not limit yourself to the functions listed here. It is possible to do much more, and the post would probably become even longer if I tried to cover everything. Use your creativity.
Points of attention
At the moment of publication, Power BI Modeling MCP Server is in Public Preview, which means it is still in a testing phase and may change significantly before the final release.
If you work in BI, it is worth following Microsoft updates and especially AI integrations such as Power BI Modeling MCP Server. Paying attention to these changes will be a competitive advantage.
The official documentation and usage examples are available in the Microsoft GitHub repository.
👇 Enjoyed this content?
Share it and keep following my publications.
Take the opportunity to read more articles!