Dashboard Widgets and Charts in Dekart

by Vladi ‐ 2 min read

Widgets and charts are live in your workspace: filter the map in the browser instead of re-querying the warehouse.

Widgets and charts are live in Dekart Cloud. Your map is now a dashboard: row counts, category charts, histograms and filters next to the map, on the same query result.

Nothing to install or upgrade. Open a map, run a query, and the charts are there.

Create a map in Dekart Cloud

What you get

Charts suggested from your query. A row count, a category chart for your first text column, and a histogram for your first number, without configuring anything.
Filters that drive the map. Click a category or brush a histogram, and the map, the counts and the other charts follow.
Charts that match the map. A category chart uses the same colors as the layer it is colored by.
Your reader needs no SQL. Share the link, they filter it themselves. The map stays connected to your database.

Why it is fast, and cheaper

In most cloud tools each map and each chart is a separate warehouse query. Filter on a category, and that is the next roundtrip to BigQuery.

Dekart uses DuckDB and SQLRooms instead:

You query BigQuery, Snowflake or Postgres once, and load the map.
Every chart is aggregated in the browser with DuckDB.
When you filter, there are no more queries.

You pay less for the data warehouse and get an instant experience.

In the video that is 81,268 Overture places from BigQuery, 254 categories, filtered down to 19,800 without a second query. Uploaded files work the same way.

Open the map from the video → BigQuery places with widgets

It works with your agent

Dekart runs with GeoSQL, the map-in-the-loop harness for Claude and Codex, so the agent can build the dashboard instead of you clicking it together. The prompt in the video:

/geosql Add 3D H3 hexagons (res 9) on top: height by place count,
color by top category. Use DuckDB. Rewire chart. Name chart, layer.

Claude aggregated the 81,268 places into 2,053 H3 cells, added the 3D layer, rewired the charts to the new aggregation and named everything for sharing.