Data is only as valuable as the ability to understand it. We are announcing Conversational Analytics in BigQuery now allows to query and talk to open-format lakehouse tables. This introduction brings enterprise-grade, agentic data analysis to your open data ecosystems, allowing you to combine the processing power of BigQuery with the immense flexibility of a lakehouse to uncover rich insights instantly with no complex SQL required.
Conversational Analytics now allows users to query open datasets using natural language prompts, providing the same BigQuery interface and workflow, expanding the execution engine to agentic data analysis that reasons across your lakehouse schema, autonomously resolving ambiguities to deliver trusted insights. This delivers the processing power of BigQuery, flexibility of a lakehouse, and the full feature set of Conversational Analytics to explore distributed, open-format data using natural language.
Whether you are querying managed Iceberg tables, cross-cloud Apache Iceberg tables (federated by Databricks Unity Catalog or external REST catalogs), or other formats supported by the Lakehouse runtime catalog, you can now do it using natural language. Once you register these tables in Lakehouse runtime catalog, Conversational Analytics acts as an unified hub for natural language insights. Additionally, as Lakehouse integrates with Knowledge Catalog, your open-format tables become discoverable alongside your standard data assets within BigQuery Studio Agent Hub.
Using Conversational Analytics for Lakehouse is intuitive and simple. When you ask a question, the agent uses metadata from the Lakehouse runtime catalog to identify table structures, column names, and data types before generating a federated SQL query. BigQuery then executes this query against the data located in your Lakehouse storage. With full access to the latest BigQuery Conversational Analytics capabilities, including disambiguation and forecasting, the agent surfaces rich insights and charts rapidly.
To get started, navigate to the Agent Hub in BigQuery Studio within the Google Cloud console to create a Data Agent or start a direct conversation. Finding an open-format table relies on the same interface you use for standard tables:
-
Search: Add a knowledge source and find your tables. You can use specific filters like TABLE_NAME catalog: CATALOG_NAME for faster discovery.
-
Select: Add your desired Lakehouse tables to your agent. You can also mix in standard BigQuery tables or object or external tables so you can chat with all your data sources at once.
-
Create: Enrich your agent’s knowledge by providing relevant business and data context.
-
Chat: Talk to the Conversational Analytics agent directly in the BigQuery UI.
-
Share: Share the agent seamlessly with other users on your team.
-
Publish: Publish and export the agent to Data Studio or Gemini Enterprise to unlock chat outside of the Cloud Console, or embed it into a custom application using the Conversational Analytics API.
To optimize performance and improve query accuracy, the platform provides configuration options for Conversational Analytics agents. Agent creators can guide agents by adding business glossaries, verified SQL queries, and custom system instructions to help the agent accurately interpret internal business and data terminology and unique Lakehouse schemas.
With Conversational Analytics in Lakehouse, organizations can analyze open data formats through a streamlined interface, making better faster business decisions across the enterprise. You can get started today. Please find documentation here. If you have questions, please reach out to bqca-feedback-external@google.com.