Avanty

Type what you want in the command panel and Avanty writes the query against the tables in your connected Metabase instance. It pulls schema metadata through the Metabase API, so column names and schema prefixes match your warehouse instead of a generic sample database. The output lands directly in the editor, which means you can run it without copying anything.

Editing is a separate flow from generating. You highlight the part of the query that needs to change, describe the adjustment and get a diff view showing what the AI SQL editor wants to replace. Accepting the change is a deliberate click, which matters when you're touching a query that already feeds a live dashboard.

The explainer breaks a long query into chunks and describes what each part does, one step at a time. This is the feature that gets the most use outside of query writing, because most analysts inherit SQL they didn't write. It won't tell you whether the logic is correct for the business question, only what the code does.

Avanty can add comments to a query and reformat it in one action. Team plans sync formatting settings across seats, so every analyst produces SQL that looks the same in review. It's a small thing until you're reading a query where every join sits on one line.

When you save a Metabase question, the extension suggests a title for the chart based on the query and the result columns. It saves the small pause of naming yet another "Count by month" card, and you can always type over it.

If a query fails, Avanty reads the database error and proposes a corrected version. The common case is a renamed or moved column after a schema change, where the fix is obvious once someone points at the right field. You still confirm the change before it replaces your query.