Messy input
Headers are wrong, date columns are inconsistent, CSV delimiters are weird, and someone pasted totals into the middle of the file.
Local-first · DuckDB-powered · Windows desktop
Clean messy Excel, CSV and Parquet files locally with DuckDB.
Bendit is a desktop workbench for importing data files, inspecting columns, writing SQL transformations, and exporting clean results — without setting up a database server or uploading your data to the cloud.
Prefer Parquet? Download the same sample as .parquet →
No signup wall
Local files stay local
SQL-first workflow
The annoying middle ground
When Excel is not enough, but a full ETL stack is overkill.
Repeatable cleanup
Use DuckDB SQL transformations instead of doing the same fragile manual Excel steps again next month.
Local by default
Keep customer files, financial exports and operational data on your own machine instead of uploading them to another SaaS tool.
Scale
Excel has hard limits. Bendit doesn't.
If you've ever hit "this workbook is too large to open" or watched Excel grind to a crawl well before a million rows, you've run into limits that are baked into the file format itself. Bendit runs on DuckDB — an embedded analytical engine built for exactly this — so instead of stopping around a million rows, it comfortably handles tens of millions on ordinary hardware.
| Feature | Excel limit | Bendit |
|---|---|---|
| Rows per worksheet | 1,048,576 | Tens of millions — bounded by your disk and RAM, not a hardcoded ceiling |
| Columns per worksheet | 16,384 (A to XFD) | No fixed column limit |
| Characters per cell | 32,767 | No fixed limit |
| Worksheet tabs per workbook | Limited by available memory | Any number of tables, limited only by disk space |
| Unique cell formats | 65,490 | Not applicable — plain typed columns, nothing per-cell to run out of |
| Hyperlinks per worksheet | 65,530 | Not applicable |
| Workbook file size | No fixed limit, but practical limits depend on RAM and Excel version | Local DuckDB files comfortably scale well past typical Excel file sizes before performance degrades |
None of this makes Excel a bad tool — it's still the fastest way to eyeball a few hundred rows. But once a file is pushing against these ceilings, or Excel just gets sluggish well before actually hitting them, that's exactly where a real query engine underneath starts to matter.
Workflow
Import → inspect → transform → export.
Bendit keeps the flow deliberately small. Open a file, inspect its columns, write SQL, preview the output and export clean CSV or Parquet.
- Import Excel, CSV and Parquet files
- Query and transform with DuckDB SQL
- Preview results in a fast grid
- Export clean output for other systems
Real examples
Built around the boring file problems that waste hours.
SQL-first
Use plain SQL to bend ugly data into shape.
No visual pipeline ceremony. No cloud job. No database server. Just a file, a query, and clean output.
See more examples →select
trim(customer_name) as customer_name,
try_cast(order_date as date) as order_date,
replace(amount, ',', '.')::decimal(18,2) as amount
from imported_excel
where customer_name is not null
order by order_date desc;Result grid
Check your work before you export it.
Every query lands in a fast, filterable grid — so you can sanity-check a transformation before it goes anywhere.
Filter without writing SQL
Every column header has its own inline filter, so you can narrow down a few million rows to the ones you care about without touching the query.
Export the view, not just the query
Active filters and sort order carry through to the export — what you see in the grid is what lands in the file, in CSV or Parquet.
Streamed straight from DuckDB
Exports run through DuckDB's own COPY, streaming the full result set directly to disk — not just whatever page happens to be loaded in the grid.
Questions people search for
Quick answers.
Can Bendit open Excel files?
Yes. Bendit is built for importing Excel files and turning messy worksheets into queryable data.
Does Bendit upload my data?
No. The product is designed around local-first file transformation. Your files stay on your machine.
Does Bendit use DuckDB?
Yes. DuckDB is the SQL engine behind Bendit’s file transformation workflow.
Is Bendit a BI tool?
No. Bendit is focused on cleaning, reshaping and exporting data files, not dashboards.
Does Bendit have Excel's 1,048,576 row limit?
No. That limit comes from the Excel file format itself. Bendit runs on DuckDB, which comfortably handles tens of millions of rows on ordinary hardware — well past where Excel would refuse to open the file at all.
Is Bendit available for macOS or Linux?
Not yet. Bendit is Windows-only today; macOS and Linux builds are on the roadmap. Email us to get notified when they're available.
Ready to test it?
Download Bendit and try it on one messy file.
Best test: take a real Excel or CSV export that normally annoys you and try cleaning it with SQL.