Klindok

Problem

Private practices in the DACH market need clinical letters (Arztbriefe) without sending patient data to a cloud service.

What I built

Klindok is offline, privacy-first AI medical documentation. It writes GDPR-compliant Arztbriefe in under 30 seconds. It runs on my own small language model.

Doctors and admins have separate access. Passwords use bcrypt. Patient data sits in a Fernet-encrypted SQLite database. Every patient-level action is written to a GDPR Art. 30 audit log.

Architecture

The app stays on the practice machine. The database schema comes first. Input is validated before it is stored. pytest covers auth and the audit log.

Results

• Under 30 seconds per Arztbrief
• 100% offline
• GDPR Art. 30 audit log of every patient-level action

What I learned

Privacy is part of the build, not a later checklist. Encryption, separate roles, and an audit log have to exist before a letter is useful to a clinic.

Stack

SQLite holds the data. Fernet encrypts it. bcrypt covers passwords. pytest checks auth and the audit log.