Your document is for your request—not our library.

What Groundquote processes

When you ask a question, Groundquote receives the document name, document text, and question. It cleans and chunks the text, retrieves relevant passages, and produces a response. The public service limits documents to 60,000 characters.

Generated answers and model providers

When AI-generated mode is configured, Groundquote sends your question and the retrieved passages—not a stored document collection—to the configured model provider. That provider handles the request under its own data and retention policies. Evidence-only mode does not make a generation request.

Feedback

A helpful or not-helpful rating is remembered in your browser. If aggregate feedback storage is enabled, Groundquote may also store an anonymous answer ID, the rating, the answer mode, and a timestamp. Document text and questions are not included in that feedback record.

Operational data

The hosting platform may process standard request information such as IP address, user agent, timing, and error logs for security and service operation. Groundquote uses an IP-derived key for rate-limiting but does not add IP addresses to its feedback table.

The public site uses Vercel Web Analytics to understand basic usage. Vercel reports those visits without cookies and does not expose an individual visitor's identity to Groundquote.

Use good judgment

The public demo is not intended for passwords, API keys, government identifiers, health records, privileged legal material, or other regulated secrets. Review your organization's policies before sending sensitive information to any hosted AI service.

Questions or security reports

The project is open source. For privacy questions, open a non-sensitive discussion on GitHub. Report suspected vulnerabilities privately through the project's security advisory form.