This post covers how you can speak to your Fabric Ontologies using voice commands. In the past I recommended chatting with your Ontologies by adding them to a Fabric data agent and then chat with your data agent in Visual Studio Code.
Now there is an ontology MCP server available. So I decided to show how you can speak to Fabric Ontologies directly within Visual Studio Code.
To manage expectations, I do not show Visual Studio Code talking back in this post. Just the text responses. I am sure there is a way you can configure this though. As always, I share links along the way.
To follow along you will need GitHub Copilot in Visual Studio Code configured.
Speak to your Fabric Ontologies
This example is based on the Ontology tutorial provided by Microsoft. After configuring the Ontology in a new workspace I was able to create the URL for my MCP server. Microsoft documents this in the “Consume Ontology as an MCP server” guide.
One key point I will highlight is that you will need to start your MCP server when following the guide.
Once the MCP server was started I asked a sample question with a prompt.
I then took it one step further and installed the VS Code Speech extension. Which added a new Start Voice chat icon.
I tested with my own voice and the results turned out well as you can see below.
However, I still had to manually click on send after speaking. So, I decided to take this one step further and configured voice typing in Windows. For a more automated experience.
I then read the prompt again and this time finished the sentence with “press enter”. I was impressed with the result.
I then tested afterwards with another prompt.
Summary
As you can see, you can speak to your Fabric Ontologies in multiple ways now. Being able to work with voice commands is great from an accessibility perspective. However, I am not going to proclaim “chat with your data is dead, long live speak with your data” just yet.
Besides, I do not want to be held responsible when people are caught talking to their laptops in public places.
Working with the ontology MCP server can bring various benefits. However, you can also work with the MCP server created by your data agent instead if you want to retain instructions in Microsoft Fabric. At the end of the day depends on your requirements.