BigQuery AI

BigQuery AI is the collection of predictive, generative, search, and assistive AI capabilities built into Google Cloud’s BigQuery data platform. Teams can train models with SQL, call Gemini and other supported models from queries, generate embeddings, run vector search, and use agents or code assistance without exporting governed warehouse data into a separate analytics product.

The capabilities are available through the Google Cloud console, SQL, the bq command-line tool, REST APIs, and notebooks. Gemini in BigQuery adds data insights, conversational analytics, data preparation, SQL and Python assistance, while BigQuery’s AI functions bring row-scale generation, classification, extraction, and enrichment into data pipelines.

Supported use cases

Enrich a large table with AI

Apply classification, summarization, extraction, or sentiment logic to warehouse rows from SQL.

Build predictive analysis in the warehouse

Train and run forecasting, anomaly detection, classification, regression, or recommendation models near the data.

Create semantic search or RAG

Generate embeddings, index them, and retrieve similar content for an application or agent.

Limitations

Edition and service dependencies

Some capabilities depend on BigQuery edition, region, Gemini or Vertex AI availability, IAM configuration, and separate model-service quotas.

Multiple billing components

BigQuery processing and storage charges can apply in addition to Vertex AI or other remote-model inference charges.