F1: from API to agent

Explore Formula 1 race data and official documents with F1 Analyst in Gemini Enterprise. Fivetran loads the data into BigQuery, and dbt prepares it for analysis and search.

flowchart LR
    O[OpenF1 APIs] --> F[Fivetran connectors]
    D[Rules, glossary, transcripts] --> F
    F --> B[BigQuery]
    B --> M[dbt models]
    M --> A[AGENTS context]
    M --> E[Embeddings + hybrid indexes]
    A --> G[F1 Analyst in Gemini Enterprise]
    E --> G

Loading

Component	Implementation
Ingestion	OpenF1 connector and documents connector.
Racing data	Normalized dbt models.
Document search	Document models compute embeddings and create hybrid indexes using native BigQuery capabilities.
Agent context	AGENTS schema describes tables, columns, joins and query guidance.
Analyst	Agent configuration and system prompt, published to Gemini Enterprise.

Setup and operation

1. Follow operations to configure credentials, IAM and connector dependencies, then deploy the connectors and complete their Fivetran syncs.
2. Configure dbt Cloud and GitHub using the deployment guide.
3. Push to main to build the dbt models, publish the AGENTS context, deploy the analyst and run evaluations.