How Can Clinics Use Jarvis Without Exposing Patient Data?

Clinics can use Jarvis AI (getjarvis.eu) safely by keeping all patient identifiers out of it, since Jarvis does not sign a HIPAA BAA and is not built to handle protected health information. Practically: de-identify before prompting, close patient records before invoking the screen-aware assistant, and connect only non-PHI accounts. Jarvis is a desktop assistant for macOS, Windows, and Linux, opened with Cmd+/ (Ctrl+/ on Windows), routing to frontier models from Anthropic, OpenAI, and Google on GDPR-aligned infrastructure with your data stored in the EU, with AES-256-GCM token encryption and no training on your data. At $16/month it accelerates clinic administration, internal documentation, supplier emails in Gmail or Outlook, and research summaries, while the certified EHR keeps the regulated record under its own BAA.

The single most important control is removing identifiers before anything reaches Jarvis. That means no names, dates of birth, medical record numbers, addresses, phone numbers, or rare diagnoses that could re-identify someone. Reframe tasks generically: instead of asking about a specific patient, ask Jarvis to help structure a generic care-instruction template or summarize a published guideline. With this habit, a large share of clinic work becomes safe to delegate, because most administrative and educational tasks never needed PHI in the first place.

Because Jarvis can read your screen when you invoke it, clinic workstations need a simple rule: minimize or close any EHR window, lab result, or patient-facing chart before pressing Cmd+/. Jarvis does not watch the screen continuously, only on invocation, but in a clinical environment that one moment of capture is exactly what you must prevent. Treat the assistant like a visitor in the room and only summon it when nothing identifiable is on display. Pair that with locked screens between patients and the exposure surface stays effectively closed.

Privacy & data