Quick Start

Five minutes with one sample file, from opening it to exporting a country summary. Instead of screenshots, each step shows the table you will be looking at.

Minute 1: open the file

Go to /app?sample=/samples/templates/ecommerce.csv, or open the app and drop any CSV, Excel, Parquet or JSON file on the page. The sample is 170 online orders. It becomes a table called ecommerce, and the grid shows all nine columns with a type icon in each header:

Types come from DuckDB reading a sample of rows. Nothing was uploaded; the badge in the header reads Connected. Your file never leaves this tab. Click a header to sort.

Minute 2: read two column cards

Hover the status header and click the small Explore & filter icon. A card opens above the grid with one bar per value:

Do the same on total. A number column gets a histogram with five figures above it:

That is already an answer to "how much did we sell": 57,311.92 across 170 orders, averaging 337.13 each.

Minute 3: filter by clicking

Double-click the delivered bar on the status card. The grid drops to 38 rows and the card shows the active filter with a clear button. Every card recalculates against the filtered rows, so the total card now reads Sum 13,914.89 and Avg 366.18. Clear the filter to get all 170 back. Filters from different cards combine with AND.

Minute 4: one transformation

In the Pipeline panel on the left, type group in Search transforms and pick Group & Aggregate. Tick country under Group by. Under Aggregations choose COUNT of order_id and SUM of total. Click Apply. A card appears in the pipeline, and the grid becomes six rows:

The card's Show SQL toggle reveals what ran:

SELECT "country", COUNT("order_id") AS "order_id_count", SUM("total") AS "total_sum" FROM "ecommerce" GROUP BY "country"

Sort by total_sum by clicking its header. In the app the sums carry a floating-point tail, 9534.050000000001 and so on; I round them with a Math step before anyone else sees them. The pencil on the card reopens the panel if you want AVG of shipping as well, and the bin icon deletes the step.

Minute 5: ask, then export

Click Ask a question above the grid and type How many rows by status? The panel answers with the counts from minute 2, the formula COUNT(*) GROUP BY status, and the SQL it ran, no API key involved. Then click Export / download and pick CSV, Excel, Parquet, JSON or one of six other formats. What downloads is the grid as it stands, every pipeline step applied.

For the same summary as charts, flip the toolbar toggle from Grid to Canvas. The Canvas reads the column roles, picks a measure and lets you break it down by country, status or month, and its Share link button packs the whole thing into a URL.