What changed
Google Cloud has announced the preview release of Data Cloud connectors for Gemini, enabling business users and analysts to query live enterprise data using natural language. The feature allows direct access to data sources such as BigQuery, operational databases like Spanner, AlloyDB, and Cloud SQL, as well as governed data from Looker. The connectors integrate with Knowledge Catalog to enforce row- and column-level security and respect end-user credentials.
What this means in practice
The new connectors provide a conversational interface for querying enterprise data without requiring users to know specific table or project names. Gemini interprets natural language questions, maps them to database structures using technical metadata and business glossaries, and executes permission-scoped queries under the user’s identity. Users can then export results directly into tools like Google Docs, Gmail, or Google Slides. The connectors operate in federated query mode, meaning data is not copied or indexed, and instead queried in place using Knowledge Catalog to locate accessible metadata and business context.
Limitations and availability
Data Cloud connectors are currently in preview and available for BigQuery, Spanner, Cloud SQL, AlloyDB for PostgreSQL, and Looker. Setup is managed through the Gemini admin console, where administrators configure default read-only connections. The connectors do not support data ingestion mode, instead focusing on federated querying. While conversational analytics agents are also available for more governed analyses, the connectors are designed for ad hoc exploration. The feature is available for Google Cloud databases and BigQuery, with guided setup and managed authentication for Google Cloud services.
