Human Rounds documentation
Human Rounds is an open-source medical platform. AI does the repetitive work around a visit — interviewing, reading paper, summarising, filing — and the care team decides. It runs today at a public hospital in Pinamar, Argentina.
Pre-consult interview
A QR at the door, the interview on the patient’s own phone.
Referral orders read by AI
A photo of the paper order, read by AI — handwriting included.
The prepared clinician
The day’s list already read: triage, reason, summary.
Scan to register
The camera reads the ID and the insurance card into the record.
Health-record handoff
Writes to the national health record, and survives it being down.
Booking
Search, real slots, reminders, cancellation and the waitlist.
How one visit runs
- Before. The patient books online, on WhatsApp, or at the desk, and answers an AI interview on their own phone.
- At the door. Reception scans the ID; a new patient is registered without retyping anything.
- In the room. The clinician opens a list that already carries a triage colour, the reason in the patient’s words, and a summary of the record.
- After. The appointment is written into the national health record, and retried on its own if that record is down.
What it is not
It is not a diagnosis engine. Every AI output is a proposal a person approves, rejects or edits, and the existing medical record stays the source of truth. Triage rules can escalate a patient, never downgrade one.
It is not measured yet. Nothing here claims a change in waiting times, absenteeism or cost — the platform has not run long enough anywhere to have that data.
Try it or run it
The live demo is a full installation of a fictional town, rebuilt every day. To put it on your own machine, start with Install. The code is on GitHub.
Next: Pre-consult interview