Open a file
Drop in CSV, Excel, or JSON. Sheets become queryable tables directly in your browser.
Nothing is uploaded Private by default · DuckDB in your browser Query spreadsheets
Query spreadsheets
without uploading them.
NoEgress turns CSV, Excel, and JSON files into a local DuckDB workspace. Write SQL, pivot and profile your data, build charts, and use optional local AI while your files stay on your device.
No file to hand? Open the live demo — sample data and a ready-made pivot, loaded locally.
No accountNo file uploadRuns locally
0 bytesuploaded by the workbench
DuckDBSQL engine in WebAssembly
CSV · XLSX · JSONquery common data files
1
2
3
4
5
6
2
3
4
5
6
SELECT
region,
SUM(revenue) AS total_revenue
FROM orders_2026
GROUP BY region
ORDER BY total_revenue DESC; Success5 rows · 18ms · DuckDB local
regiontotal_revenueorders
North America$1,284,9208,412
Europe$948,2106,801
Asia Pacific$734,8805,294
100% localYour files never leave
Query complete24,891 rows in 18ms
A simpler data workflow
From raw spreadsheet to
clear answer in minutes.
No setup scripts, database server, or cloud pipeline. Open a file and work with it immediately.
Query with confidence
Write SQL, drag a pivot together, or add columns and filters as recipe steps — no SQL required.
You review before runningUnderstand and export
Profile columns, chart results, compare versions, and export to CSV, Excel, or Parquet.
Deterministic local resultsBuilt for serious exploration
A focused workbench,
not another dashboard toy.
Everything needed to move from an unfamiliar dataset to a reproducible result.
Real SQL. Real execution.
DuckDB runs queries against your actual data instead of an AI-generated approximation.
SELECT region, SUM(revenue)FROM orders GROUP BY region; Completed locally in 18ms
Optional local AI
Generate, explain, repair, and optimize SQL through WebGPU models cached on your device.
Schema-aware editor
Autocomplete, formatting, query history, saved queries, a command palette, and undo/redo throughout.
Fast table inspection
Search, sort, and inspect large result sets with virtualized rendering — switching between tables stays instant, even with large files.
Useful exports
Take results with you as CSV, Excel, copied tabular data, or locally generated Parquet.
Chat with your data
Ask questions in plain language and get schema-grounded answers from a fully local model — it suggests SQL instead of guessing numbers.
Instant chart suggestions
Upload a file and immediately see auto-picked bar, line, pie, and scatter charts based on your column types — no manual setup.
Pivot tables, no spreadsheet
Drag fields into rows, columns, values, and filters. Six aggregations, subtotals and grand totals, in-cell sparkline trends, and colourblind-safe heatmap shading — compiled to DuckDB SQL you can inspect.
Profile before you trust
Every column gets missing-value counts, duplicates, type violations, min/max, median, mode, and a distribution — so data quality problems surface before your conclusions do.
Transform without SQL
Build derived columns, filters, renames, and type casts as ordered recipe steps. Each step is a real query, reversible with undo.
Join and stack tables
Append files with matching columns, or join two tables on a key — with suggested join keys when a local model is loaded.
Diff two versions
Match last month's export against this month's on any set of key columns and see exactly what was added, removed, and changed — NULLs included.
Reusable recipes and workflows
Chain multi-step SQL workflows, then save recipes, workflows, and pivot layouts as small JSON files to re-run against next week's data.
Privacy is the architecture
Your dataset stays
where it belongs.
NoEgress is designed around browser-native execution. File parsing, SQL, charts, and supported AI workflows happen on your machine.
- ✓ No file upload endpoint
- ✓ DuckDB runs in-browser with WebAssembly
- ✓ AI models run locally through WebGPU
- ✓ Queries remain visible and reviewable
- ✓ Datasets persist locally via OPFS — no re-upload after reload
- ✓ Installable, and works offline after the first load
Questions
Running SQL locally,
explained.
The things people ask before opening their first file.
Can I run SQL on a CSV file without uploading it?
Yes. NoEgress loads your CSV, Excel, or JSON file into DuckDB compiled to WebAssembly, which runs inside your browser tab. The file is read from your disk by the browser and never sent to a server — there is no upload endpoint to send it to.
Do I need an account to use NoEgress?
No. There is no sign-up, no login, and no email required. Open the workspace, pick a file, and start querying.
How large a file can it handle?
DuckDB-WASM works comfortably with files in the hundreds of megabytes on a typical laptop. The practical ceiling is your browser tab memory rather than a server quota, and columnar execution means wide files stay fast.
Which file formats are supported?
CSV and other delimited text files, Excel workbooks (.xlsx and .xls), and JSON. Results can be exported back out as CSV or Excel.
Is the AI assistant sending my data to OpenAI or another provider?
No. The optional assistant runs a small open-weight model locally through WebGPU, downloaded once to your browser cache. Prompts, schema, and results stay on your device, and the feature is entirely optional — the SQL workbench works without it.
Does it work offline?
Yes. NoEgress is an installable progressive web app; after the first visit the workbench loads and runs with no network connection.
How is this different from an online CSV or SQL tool?
Hosted tools copy your file onto their servers to query it, which makes them unusable for confidential or regulated data. NoEgress moves the query engine to the data instead: the same SQL, pivots, and charts, with nothing leaving the machine you are sitting at.
Ready when your file isOpen a spreadsheet.
Open a spreadsheet.
Start asking better questions.
Turn local files into a queryable DuckDB workspace without creating an account or uploading data.
Open NoEgress