Privacy

ParquetView is built so that opening a local file does not upload it to parquetview.com. There is no ParquetView account and no ParquetView file store.

What stays in the browser

• File bytes you drop onto the page, read via the browser File API.
• SQL you run, filters, charts, and exports — computed with DuckDB WASM on this device.
• A copy of files in Origin Private File System so they can reload on the next visit. Use “Clear all local data” in the viewer to wipe them.

What the site itself loads

The viewer downloads JavaScript, CSS, fonts (Google Fonts), and the DuckDB WebAssembly binary from our host (Azure Static Web Apps) and from Google Fonts. DuckDB may fetch extensions from extensions.duckdb.org. Those requests are engine code, not your dataset.

Optional features that do use the network

• Azure Blob — if you connect a storage account, your browser talks to Microsoft login and your blobs. ParquetView does not proxy the file through our API.
• Gemini “Ask your data” — only if you paste your own API key. The prompt (and therefore anything you typed from the data) is sent to Google Generative Language from your browser.
• Umami — cookieless page views and a few named product events (file opened, export, AI SQL). No cookies, no file names, no SQL text, and no account. Hosted by Umami Cloud. Visitors who send a Do Not Track header are not counted.
• Application Insights — if the deployed build has a connection string, we send opt-in, redacted events. Automatic route, AJAX, and exception tracking are off so file names and SQL are not vacuumed up.

Cookies and accounts

No product cookie wall, no signup. Theme and query history use local storage. Umami does not set a cookie.

Contact

Questions: contact@parquetview.com.