BigQuery Graph is now GA: the knowledge foundation for the agentic era
Google Cloud has made BigQuery Graph generally available, integrating native graph analytics into its data warehouse to connect enterprise data without ETL or silos.
BigQuery Graph, now generally available, embeds graph capabilities directly into BigQuery, eliminating the need to export data to separate graph databases. The service supports ISO-standard GQL alongside SQL, enabling native traversals without ETL. Built on BigQuery’s infrastructure, it scales to petabyte-level data while maintaining row- and column-level security, and integrates with BigQuery ML and AI functions within the same query environment.
The release introduces performance improvements, including faster path-finding for acyclic and undirected traversals—GQL is twice as fast since preview, with undirected traversal 100 times faster—and lower query latency for interactive access. New features like the CALL statement and extended subquery support allow analysts to write modular queries, which AI agents can reuse as tools, bridging graph and relational analytics.
BigQuery Graph supports cross-cloud virtual knowledge graphs, enabling agents to traverse data across Google Cloud, AWS, and other environments without moving data. This capability allows support agents, for example, to query supplier and customer records spanning multiple clouds in a single traversal, reducing the need for per-request data stitching.
The service also introduces conversational analytics, allowing users to query graphs in natural language through BigQuery’s interface. Agents can interact with the graph via an MCP server or publish conversational data agents directly. Google has packaged graph expertise into an agent skill, available in tools like Antigravity and Visual Studio Code, to automate graph modeling and querying for enterprise datasets.