Building Reliable Conversational Agents with Genie

This course teaches you how to design, build, and maintain a Databricks Genie Space, a natural language interface that enables business users to ask questions about governed data and receive SQL-backed answers without writing code.

You will learn how Genie fits into the Databricks AI/BI product family and how it translates natural language into reliable SQL queries. The course focuses on what it takes to create a Genie Space that delivers accurate, consistent, and trustworthy results.

You will follow a complete end-to-end workflow, from understanding source data and defining benchmarks to configuring and refining a Genie Space using the full set of Knowledge Store curation tools. These include metadata, synonyms, prompt matching, SQL logic, example queries, and text instructions.

You will also learn how to share Genie Spaces through Databricks One, understand how Unity Catalog governance is automatically enforced, and use monitoring and user feedback to continuously improve quality over time.

By the end of the course, you will be able to create and manage a production-ready Genie Space that delivers governed, self-service conversational analytics at scale.

Prerequisites

General familiarity with the Databricks workspace - You can navigate the workspace UI, attach compute resources, and run notebooks.

Familiarity with Unity Catalog concepts - You understand catalogs, schemas, and tables in Unity Catalog. Prior experience with permissions, tags, or column masks is helpful but not required.

Intermediate SQL proficiency - You can read and write SQL queries including SELECT, JOIN, GROUP BY, and aggregate functions.

Access to a Databricks workspace with the Databricks SQL entitlement and CAN USE permission on a serverless SQL warehouse.Note: Hands-on training courses will be updated to reflect the newest product and feature announcements from Data + AI Summit in June 2026.