Data Workshop is powered by an embedded columnar engine (DuckDB), so aggregate queries over millions of rows return in milliseconds without a warehouse to provision. It lives at Data Workshop in the dashboard.
Datasets from files
Create a dataset and upload one or more .csv, .parquet, or .json files; each becomes a table you can query. Columns and types are inferred on ingest. Storage per account is capped by plan (Plans & limits), and uploads pause when you approach the cap rather than failing mid-write.
-- query an uploaded dataset SELECT country, count(*) AS orders, sum(total) AS revenue FROM orders GROUP BY country ORDER BY revenue DESC LIMIT 20;
Your project databases
Data Workshop can also open the Postgres database attached to any project you own — browse tables, inspect schemas, and run SQL against it (read-only by default; toggle on to write) without leaving the dashboard. The connection is server-side; no credentials are exposed to the browser.
Cross-database queries, on demand or on a schedule
To read or write across your databases — join two projects, or refresh a rollup on a cadence — use scheduled SQL queries. They reference any database your account owns by name and run isolated in Montreal, writing results straight into the project Postgres you name.
Saved queries, export & cancel
Save a query you rerun often, export any result set to CSV, and cancel a long-running statement from the query page. Daily query volume is plan-tiered; heavy or runaway queries can be cancelled rather than left to consume the day’s budget.
Privacy
Everything runs on Canner’s Canadian infrastructure. Uploaded data and query results stay within your account and are never sent to a third-party analytics service — the same data-residency posture as the rest of the platform. See data sovereignty for the full picture.