Google BigQuery

BigQuery is the serverless data warehouse most modern Google Cloud stacks default to — partitioned tables, nested fields, cross-project queries, all at petabyte scale. myai treats it as a first-class data source: ask a question in chat, get a table back; schedule a function, post the result to a Slack channel; build a canvas, mix BigQuery with anything else.

What you can do

• Pull live BigQuery data into a chat — "show me yesterday's order volume by region" returns a table built directly from your warehouse, no BigQuery console.
• Run cross-project queries when the data lives in one project and the analytics happens in another.
• Functions that read BigQuery on a schedule — daily rollups, weekly anomaly checks, monthly reporting digests, posted to wherever the team reads them.
• Cross-source canvases mixing BigQuery with Postgres, Snowflake, or Fabric data in a single view.
• Nested/repeated field handling for the kind of analytical schemas BigQuery is best at.

How to connect

Uses a Google Cloud service account JSON key scoped to the BigQuery datasets you want exposed. The minimum read-only pair of IAM roles is BigQuery Data Viewer (read schema + data) and BigQuery Job User (submit query jobs — required even for SELECT). Add BigQuery Data Editor if you need to write tables. Avoid BigQuery Admin unless you specifically need it — it grants dataset and job ACL management on top.

If the service account lives in a different GCP project than the datasets being queried, the IAM roles aren't enough on their own — each dataset has to be shared with the service account email from the BigQuery console (Dataset → Sharing → Add principal). This is the most common stumble.

Connect Google BigQuery in myai

Common use cases

• Executive metrics dashboards rolled up from BigQuery into a single canvas — refresh and drill into anomalies in the same conversation.
• Customer-facing reports powered by BigQuery without standing up a separate analytics seat for each customer.
• Daily ops digests — a scheduled function reads the previous day's records, summarizes them, and posts to the team's chat.

Reference

• Connection model: GCP service account JSON key
• Minimum IAM: BigQuery Data Viewer + BigQuery Job User
• Cross-project access requires explicit dataset sharing
• Scope: ~8 operations spanning query execution, table introspection, and dataset metadata