Consilium

Privacy policy

Last updated: October 2026

Draft — this document will be reviewed by a privacy lawyer before the service opens to the public.

Medical information is among the most sensitive data there is. This is how we treat it: we collect only what is needed, encrypt it, never share it, and let you see and delete everything at any moment.

What we store

  • Medical documents you upload (PDFs and images), free text and intake answers, the case summary the system extracts, and the discussions and reports.
  • The email address you sign in with (online version).
  • An activity log (audit log): when a file was opened, a document uploaded, a discussion run, data exported or deleted — identifiers and counts only, no names or medical content.
  • We do not collect data for advertising, and we do not sell or share data with third parties for commercial purposes.

What it is used for

  • Only to produce your discussion and report, and to let later discussions build on the same patient's history.
  • Your data is not used to train models.

Where it is stored and how it is protected

  • Every document, case summary and discussion is encrypted with AES-256 before it is stored. Without the encryption key the data is unreadable — including to the storage provider.
  • Online version: the site is hosted on Render (Frankfurt) and the encrypted data is kept in private Supabase storage. Each file belongs only to the user who created it, and every access is checked on the server.
  • Local version: data stays on your own computer.
  • The connection is encrypted (HTTPS), requests from other sites are blocked, and medical data is not kept in the browser cache.
  • Names and ID numbers are not passed to the virtual specialists; ID and phone numbers are automatically removed from text sent to the discussion.

Processing by third parties

  • Reading documents and running the discussion uses Anthropic's Claude models, so document content is sent to Anthropic for processing.
  • Under Anthropic's terms, data sent through the API is not used to train models by default. Before opening to the public we will sign a data processing agreement and request zero data retention.
  • Optional, only if explicitly enabled: for an additional independent opinion, the question and case summary (without names or ID numbers) are also sent to OpenAI (ChatGPT) and Google (Gemini).
  • Research searches run against public databases (PubMed, ClinicalTrials.gov, FDA) — only general medical search terms are sent, never identifying details.

Your rights

  • Access: you can download everything stored about a file at any time.
  • Correction: you can correct the case summary at any stage; your corrections take precedence over what was extracted.
  • Deletion: on the "My data" page you can permanently delete a document, a discussion, a whole file — or your account and all data.

Retention and deletion

  • Data is kept until you delete it. We recommend deleting files you no longer use.
  • An automatic deletion policy for inactive files will be defined and published here in advance.