Skip to content

Demo: Configure the Microsoft Fabric MCP Connector in Copilot Studio

List of References (Click to expand)
Table of Content (Click to expand)

How we move from basic coding all the way to AI agents?

flowchart LR A[Scripting: Line-by-line instructions] --> B[Machine Learning: Packages + statistical foundations] B --> C[LLMs: Reasoning, understanding, human-like responses] C --> D[Agents: LLMs with ability to act] %% Styling classDef step fill:#4a90e2,stroke:#333,stroke-width:2px,color:#fff,font-weight:bold; class A,B,C,D step;
More details about it here (Click to expand)
  • We all start with scripting, no matter the language, it’s the first step. Simple/complex instructions, written line by line, to get something done
  • Then comes machine learning. At this stage, we’re not reinventing the math, we’re leveraging powerful packages built on deep statistical and mathematical foundations. These tools let us automate smarter processes, like reviewing claims with predictive analytics. You’re not just coding anymore; you’re building systems that learn and adapt.
  • LLMs. This is what most people mean when they say AI. Think of yourself as the architect, and the LLM as your strategic engine. You can plug into it via an API, a key, or through integrated services. It’s not just about automation, it’s about reasoning, understanding, and generating human-like responses.
  • And finally, agents. These are LLMs with the ability to act. They don’t just respond, they take initiative. They can create code, trigger workflows, make decisions, interact with tools, with other agents. It’s where intelligence meets execution
Before Fabric (Click to expand)

E.g of a solution prior Microsoft Fabric:

Centered Image

Now from one place:

Centered Image
Centered Image

From Microsoft Documentation

Low code approach (Copilot Studio Agents):

MCP-Fabric-Agent2Agent-CopilotStudio-databricks+data-agents+copilot-Studio drawio

Full code approach (AI Foundry Agents):

MCP-Fabric-Agent2Agent-CopilotStudio-data agent Fabric + AI Foundry drawio

Note About the licensing: Microsoft 365 Copilot Pricing – AI Agents | Copilot Studio
Copilot Credit consumption rates: - Regular (non-generative AI) = 1 Copilot Credit; and - Generative AI (answers over your data) = 2 Copilot Credits. Customers can use a mix of classic and generative AI messages with the capacity. Learn more - Copilot Credits are offered via prepaid Copilot Credit pack subscription licenses. One Copilot Credit pack = 25,000 Copilot Credits. - The Copilot Studio pay-as-you-go meter allows you to post-pay based on the actual number of Copilot Credits consumed in a month.

image

Prepare your data

Medallion Architecture: Explore the structured approach to data management.

Centered Image

Or if you need to handle complex data transformations and large-scale data processing, you can use our combined solution of Fabric + Databricks. This powerful combination leverages the strengths of both platforms to provide a robust data processing pipeline. This workshop on Fabric with Databricks for Data Analytics offers a comprehensive step-by-step guide on developing Medallion Architecture using Fabric and Databricks.

image

Click here to read more about Azure Databricks + Fabric better together

Use Fabric Data Agent (Preferred for Semantic Models)

Data Agent (AI skills in your data) in Microsoft Fabric enable users to create conversational AI experiences that answer questions about data stored in lakehouses, warehouses, Power BI semantic models, and KQL databases. These skills make data insights accessible and actionable, allowing users to interact with data naturally and receive relevant answers without needing technical expertise. You can create custom Q&A systems using generative AI, guiding the AI with instructions and examples to ensure it understands your organization's context and data.

Key Features:

  • Customizable Q&A Systems: Tailor the AI to answer specific questions relevant to your organization.
  • Generative AI: Leverage advanced AI to interact with your data, enhancing data-driven decision-making.
  • Ease of Use: Once set up, users can simply ask questions and get accurate answers without needing deep technical knowledge.

E.g:

image

image

Setup required (Click to expand)
  1. Please ensure you read all the prerequisites

  2. Tenant switch enabled: These features must be activated as mentioned here prerequisites

  3. F2 Fabric capacity or higher: Ensure you have the appropriate Fabric capacity.

  4. Workspace configured with Fabric Capacity:

    image

    image

  5. At least one of these: A warehouse, a lakehouse, one or more Power BI semantic models, or a KQL database with data.

How it works (Click to expand)

  1. Go to the workspace, click on + New item, search for Data agent, and select it.

    image

  2. Choose the name for the Data agent instance:

  3. Add data:

  4. Relate tables, and start asking!

    image

Note Or use: MCP Server (For Custom Integrations) you create it and connect via Power Platform for example.

Examples of what to ask (Click to expand)
Question Example of it looks
Data Exploration
- Can you provide an overview of this dataset?
- Are there any missing values or anomalies in this dataset?		 image
What are the key variables in this data? image
Data Insights
What patterns or trends can you identify in this data? image
Can you highlight any correlations between variables? image
What are the outliers in this dataset? image

Fabric Data linked to Copilot Studio (MCP)

  1. Create a Data Agent in Microsoft Fabric and connect it to your semantic model.
  2. Publish the agent and verify that it is running successfully.
  3. In Copilot Studio, go to Agents → Add an agent, then select your Fabric Data Agent from the list.
  4. Test the integration by running sample queries such as: What was the revenue last quarter?

image

image