A real database in the page, not a spreadsheet script
The files people most need to analyse — customer lists, sales exports, payroll, patient records, survey responses — are the ones they should least be uploading to a website they found five minutes ago. And the usual alternatives each fall short somewhere: spreadsheets slow to a crawl past a few hundred thousand rows, notebooks need Python and an environment, and BI tools want an account, a connector and a subscription.
These tools run DuckDB, an analytical SQL database built for exactly this kind of work, compiled to WebAssembly and loaded into the page. It reads your file straight from your disk, keeps its working copy in the tab's memory, and answers in milliseconds on files with millions of rows. Nothing is sent to a server, because there is no server doing the work.
Which data tool to reach for
For anything exploratory, open Data Studio: filter, clean, group, pivot and join with the toolbar, chart the result, and write SQL when you want to — every change is a step you can remove. The single-purpose tools are quicker when you know exactly what you need.
- Two versions of a file: Compare CSV Files lists added, removed and changed rows
- A folder of exports with the same columns: Merge CSV Files
- A mailing list with repeats: Remove Duplicates, by whole row or by email
- A file too big for an import form or for Excel: Split CSV File
- Format conversions: Excel to CSV, CSV to Excel and the Parquet viewer
What running in a browser costs, and what it buys
A browser tab has less memory to work with than a database server, so there is a ceiling: files of a few hundred megabytes are comfortable, and larger ones are read from disk on each query rather than held in memory, which is slower. For the everyday work of cleaning, summarising and reconciling exports, that ceiling is rarely reached — and in exchange, the data never moves.
Browse related how-tos in our Guides, or read more about why Whiztools keeps processing on-device in About.