Fabric Data Agents enable users to interact with their data through natural language across multiple sources, including Lakehouses, Warehouses, SQL and KQL databases, Power BI semantic models, Graph, ontologies, and content indexed through Azure AI Search.

In this session, we will explore how the agent uses the user’s credentials to access only authorized data, identify the most appropriate source, interpret the question, and translate it into the corresponding query.

We will also demonstrate how to enrich a Data Agent with targeted instructions, descriptions, and example queries to guide its behavior and improve the quality and accuracy of its answers.

Finally, we will address one of the product’s current limitations. Although Microsoft documentation states that English is the only officially supported language, we will show how to configure a Fabric Data Agent so that it can understand questions and return answers in Italian.

An end-to-end demo will complete the journey, from configuring the data sources to generating a ready-to-use answer in Italian.