Privacy

Last updated: 13 November 2026.

In short: Quantilo does not collect your data. Your files are read and analysed on your own device: inside your browser in the web version, on your computer in the app, Docker image and Python library. No account is needed (signing in with Google on the web version is optional), and there is no tracking cookie, no analytics and no advertising. Only two optional features send anything out, and only when you use them: the AI assistant of the app on your computer (after you add an API key), and signing in with Google on the web version (to Google, never your data).

Who is responsible

Quantilo is published by Cyril KFOURI, a private individual (see the Legal notice). For any question about this page, open an issue on GitHub or write to cyrilkfouri@gmail.com.

The web version (quantilo.cyrilkfouri.com)

• Your files are read by Python running inside your browser tab (Pyodide). They are never uploaded: neither the publisher nor the host receives them.
• Kept on your device: the datasets you import, their results and the app settings, in this site's browser storage (IndexedDB), so they are still there on your next visit; display preferences (theme, density, sidebar, page sizes) in localStorage. Nothing of this is sent anywhere. Delete a dataset in the app, or clear this site's data in your browser, to remove them.
• Libraries downloaded at start: the page loads Python and its scientific libraries from the jsDelivr CDN (cdn.jsdelivr.net). As with any download, jsDelivr sees your IP address and browser; none of your data is sent. See the jsDelivr privacy policy.
• Hosting: the site is served by Vercel Inc. Like every web host, Vercel processes the technical data needed to deliver a page (IP address, address requested, date, browser) and keeps short-lived logs for security, under its own privacy policy. The publisher does not use these logs to identify visitors.
• Signing in is optional. The first visit offers Continue as guest and Sign in with Google. As a guest, nothing is sent to Google or to anyone else: Google's sign-in script is not even loaded. If you click Sign in with Google, the page then loads Google Identity Services from accounts.google.com; Google receives what any sign-in needs (your IP address, your browser, the site you sign in to, and your Google account if you pick one) and processes it under the Google privacy policy. Google answers with a signed token naming you; the page reads from it your name, e-mail address and profile picture, keeps them with the time of sign-in in this site's localStorage (in your browser only), and throws the token away: it is sent to no server, the publisher receives nothing. Your picture is then loaded from Google's image servers (googleusercontent.com), and Google's script may set a small g_state cookie on this site to remember the state of its sign-in prompt: the only cookie of the site, and only after you asked to sign in. Signing in unlocks nothing for now. Sign out (Settings › Account) deletes the name, e-mail and picture from your browser; clearing this site's data does too. To remove Quantilo from your Google account, see your Google account's connections.
• No tracking: no tracking cookie, no analytics or audience measurement, no advertising, no fingerprinting, no third-party embed (Google's sign-in button appears only after you ask for it). That is why there is no cookie banner: the only storage is what the app needs to work and to remember your display choices.
• No AI: the AI assistant is not part of the web version.

The app on your computer (local app, Docker) and the Python library

• Everything runs on your machine. Imported files are copied into a local workspace folder; the original is never modified. The app only listens on your own computer (127.0.0.1) and refuses requests coming from other websites.
• Nothing is sent to the publisher: no telemetry, no usage statistics, no automatic update check.
• The network is used only when you ask for it: to install Quantilo and its packages (GitHub, PyPI, Docker Hub), and by the AI assistant below.

The AI assistant (optional, app on your computer only)

• It is off until you add an API key from a provider you choose: Anthropic (Claude), Google Gemini, OpenAI, OpenRouter, Groq, Mistral, DeepSeek, or an OpenAI-compatible server (for example a local Ollama, in which case nothing leaves your computer).
• What is sent, and only when you send a message: your messages, the dataset's name, its column names and types, and the results of the analyses the assistant runs (counts, means, test results, most frequent values). Those results can contain values from your data, such as the most frequent categories. Whole rows are never sent unless you turn on row sharing in the assistant settings (off by default): then up to 20 rows at a time, after you allow each request if you chose Ask each time, with identifiers, e-mails, phones, addresses and postcodes masked unless you unmask a column.
• To whom: directly from your computer to the provider you configured. The publisher receives nothing. The provider processes the data under its own terms and privacy policy and may keep it for a while: read them before sending confidential data.
• Your API key stays on your computer (an environment variable, or a file in your workspace). The app never shows it in full, never logs it and never sends it anywhere but to its provider.
• AI can be wrong. The answers are written by an AI model; every figure it gives links to the analysis it comes from, so you can check it.

Your rights

Under the GDPR you have the right to access, correct and erase your personal data, and to object to its use. Because the publisher holds no personal data about you, there is usually nothing to hand over or delete; the data on your device is under your control. For the host's logs, contact Vercel; for the AI assistant, contact the provider you chose. You can also complain to the French data protection authority, the CNIL.

California residents: the publisher does not sell or share personal information.

Changes

This page is dated. Every change is kept in the project's public history on GitHub.