Asgard AI

Google BigQuery

Google

System integration

Lets the Agent browse each end user's own BigQuery datasets and run read-only SQL queries on them, with a cap on how much data one query can scan.

Connection routes

About the routes →
  • Connector

    OAuth authorization

    Each end user connects their own Google account once. The Agent sees and queries only the BigQuery data that person can see, and the queries run in that person's name. Needs Google Workspace.

Documentation

Lets Asgard Agents work with Google BigQuery for each end user (e.g. members of the organization):

  • See which projects, datasets and tables that person can open
  • Read the columns and descriptions of a table
  • Run a read-only SQL query and answer from the result

The Agent only reads. It does not create, change or delete tables or data. Before each query, the Agent checks how much data the query will scan, and it does not run a query over the limit set for the deployment.

What you need

  • Google Workspace for the organization.
  • A Google Cloud project with the BigQuery API turned on, and an OAuth client in it. If there isn't one, the FDE Agent walks you through creating it step by step (about 15 minutes). The client can be set to Internal, with no Google verification.
  • Each end user needs BigQuery access in Google Cloud: permission to run queries in one project, and permission to read the datasets. An administrator of the Google Cloud project gives this access.
  • BigQuery charges the project that runs the query, by the amount of data scanned.

Rollout consultation

Google and Google BigQuery are trademarks of their owner.