Transformations & queries

Raw datasets are rarely the shape you want to analyze. Transformations turn them into clean tables; queries turn those into answers and charts.

Transformations

A transformation is SQL that produces a named table. Create one from Transformations → New: give it a name, write a SELECT over your datasets, and run it to materialize the output.

The transformation and its output are one artifact on one page: the result rows, the schema, and the SQL recipe that produced them — nothing hides in a pipeline somewhere else. When an input dataset changes, the transformation is marked stale so you know its output no longer reflects the current data; re-run it to refresh.

The Lineage page draws the whole graph — sources → datasets → transformations → queries — and can run the entire pipeline in dependency order, upstream tables first, so one pass leaves everything fresh.

The SQL editor

SQL opens a scratch editor for exploration: the schema browser alongside shows every table and column, and ⌘⏎ runs the statement. Queries are read-only by design — you can never damage the data from the editor. Results export to CSV when you need them elsewhere.

Saved queries

When a result is worth keeping, save it as a query: the SQL plus a chart configuration. Saved queries re-run when opened, so what you see is always the current data — never a stale export. Charts cover the common shapes (bar, line, area, pie, scatter, and more), and the underlying rows stay one click away.

Faster with an agent — describe the table you want: "build a transformation that joins funding rounds to companies and normalizes amounts to EUR millions, then save a query charting funding by segment." The agent writes the SQL, materializes it, saves the chart, and you review the finished artifacts in the UI.