Security & trust

Built to be checked.

Your data stays yours

Each workspace is isolated at the database level: every query runs inside the workspace that asked, and a missing context returns nothing. Case content is shared per person, not by role, so an admin does not automatically see a case.

You approve what we read

Uploads go to private storage and are checked for type, active content, credentials and personal data before anything reads them. Nothing is analysed until you review and approve it, under a data policy you control.

Strong sign-in

Two-step verification (authenticator app) is required for owners and admins. Sessions are server-side, revocable, and bound to a CSRF proof. Passwords and one-time codes go to the identity provider, never to Ringo.

Staff access is narrow and visible

Ringo staff see workspace metadata only. Support access to a case needs your authorisation and a second staff approval, expires in under an hour, and shows a summary, not content. As a last resort the platform owner (super admin) can look at the product exactly as a given person sees it: read-only, for at most an hour, with a written reason, a fresh sign-in and two-step verification, and a record in the audit trail of every workspace that person belongs to, where your owners can see it. Every staff action needs a written reason and is audited.

Everything is traceable

Each finding states what it rests on: your evidence (with the quoted text), the map you described, or an explicit “unknown”. Dossiers are immutable once published; corrections are new versions or visible withdrawals.

Deletion that means it

You can delete a case or workspace. After a short grace period live data and files are removed. Backups expire on a fixed schedule, which we state rather than hide.

Honest limits (pilot)

  • Analyses always start with rules that read your map and evidence. Optionally, language models (through OpenRouter) add findings and a summary, but only for cases whose data policy allows it and only on evidence you approved for it. Every model statement is checked against your own material before it is kept, and each dossier says which route and models produced it.
  • Uploads get type, active-content and sensitive-data checks and the standard virus-test signature. A full antivirus engine (ClamAV) is supported but not connected yet.
  • Payments are a sandbox. Invitation emails are sent only once the operator has connected and verified an email provider; until then you share invitation links yourself.
  • Data is stored in Google Cloud, region europe-west1 (Belgium).
  • Legal documents are pilot drafts pending counsel review.

Found a vulnerability? Please tell us through the contact form and mark it as security. We will respond promptly.

Last updated: October 2026