What is a data agent?

Summary

• A data agent is a conversational AI system that lets people ask questions of governed enterprise data in everyday language and get trusted answers — no SQL and no analyst required.
• On Databricks, data agents are delivered as AI/BI Genie (Genie Agents, formerly Genie Agents): domain-specific agents grounded in your data and business terminology.
• How it works: Genie translates a natural-language question into SQL and returns the query, a results table, and auto-generated visualizations, asking clarifying questions when intent is ambiguous.
• Governed by Unity Catalog: users see only data they are authorized to access, every query is read-only, and all interactions are logged for audit.
• Curated by domain experts with datasets, example questions, sample SQL, business vocabulary, and trusted functions so answers stay accurate in your organization's context.

What is a data agent?

A data agent is a conversational AI system that lets business users ask questions of governed enterprise data in everyday language and receive trusted answers — including a results table, visualizations, and an explanation of how the answer was reached — without writing SQL or waiting on an analyst. On Databricks, data agents are delivered as AI/BI Genie, also called Genie Agents (formerly Genie Agents): domain-specific agents that bridge general AI capabilities and an organization's specific data context and business terminology.

Why Databricks AI/BI Genie as a data agent

• Natural language to SQL. Genie selects the relevant table and column names and descriptions to translate a business question into an equivalent SQL query, then responds with the generated query, a results table, and auto-generated visualizations.
• Clarifies ambiguous intent. When Genie cannot confidently generate an answer, it asks follow-up questions to clarify the user's intent before responding.
• Governed access through Unity Catalog. Every interaction runs within Unity Catalog access policies: users see only the data they are authorized to access, every query is read-only, and all interactions are logged for audit.
• Grounded in your business context. Genie is tailored to your organization's business terminology and data semantics, so answers are relevant and consistent rather than generic, and it can return certified answers for query patterns that data teams specify.
• Self-service beyond dashboards. Business users can ask open-ended questions that go past existing dashboards and reports without advanced technical expertise, and Genie surfaces drivers, anomalies, and relationships along with the answer.
• Improves over time. Genie learns your data and semantics continuously from human feedback and ongoing curator refinement.

Getting started

• Curate a Genie Agent: register datasets (up to 30 tables or views) to Unity Catalog, add sample SQL and example questions, define business vocabulary and metric expressions, and add trusted functions for validated logic.
• Follow Genie Agents best practices to make answers accurate and reliable.
• Read AI/BI Genie is now generally available and Pushing the frontier of data agents with Genie.

FAQs

What is a data agent?

A data agent is a conversational AI interface that answers natural-language questions over governed data by translating them into SQL and returning results, visualizations, and explanations. On Databricks, this is AI/BI Genie.

How is a data agent governed?

On Databricks, data agents run within Unity Catalog access policies: users see only the data they are authorized to access, every query is read-only, and every interaction is logged for audit.

How is a data agent set up?

Domain experts curate it by registering up to 30 tables or views, adding sample SQL, example questions, business vocabulary and metric definitions, and trusted functions, so the agent answers accurately within the organization's context.