Human Rounds

The open-source, AI-native medical platform

Humans do the rounds. AI does the rest.

AI interviews patients before the visit, reads handwritten referral orders from a photo, judges urgency, and writes to the health record. The care team walks in prepared. Live today in a real public hospital. Free to run, open to inspect.

Live demo Open source Documentation

Your health network

Who it talks to

Every patient encounter, prepared by AI.

Human Rounds automates the repetitive work before, during, and after each visit. Your existing medical record remains the source of truth — clinicians review every recommendation.

  1. 01
    Before the visit

    AI interviews and prioritizes patients

    Patients describe their problem in their own words. AI asks follow-up questions, estimates urgency, and places the patient in the right queue.

  2. 02
    During the visit

    Everything the clinician needs is already prepared

    The dashboard surfaces relevant history, medications, past visits, trends, and key findings before the conversation begins.

  3. 03
    Around the visit

    One workflow instead of disconnected systems

    Registration, scheduling, arrivals, waiting-room management, queues, and insurance claims run in one coordinated workflow.

  4. 04
    After the visit

    Documentation completes itself

    The consultation is summarized automatically. Prescriptions, referrals, lab orders, and follow-up instructions are generated for clinician review.

Roadmap

Planned

2

Not started yet

  • Clinical-history summary

    Reads today's symptoms, finds what matters in years of history, and charts it as data-dense, Tufte-style graphics.

  • More regional connectors

    Identity, coverage and record systems beyond Argentina

Prototyped

5

Prototype ready, pending Human Rounds review

  • SOAP summary + differential probabilities

    The dictated interview arrives structured: SOAP note, ranked differential, draft orders and prescriptions to review

  • Voice intake

    The patient dictates symptoms and answers follow-ups

  • Scan-to-register

    DNI + insurance credential scan fills the health-record registration

  • Triage codes

    With deterministic red-flag rules the model can escalate but never downgrade

  • Orders by checkbox

    To the electronic system or the printer

Live

12

Running at the Pinamar reference installation

Use cases

Three letters to a cardiologist

Grace needs to see a cardiologist. She doesn't know which health centre has one, or when.

The doctor already knows the case

Dr. Nguyen opens today's list before the first patient arrives. Each row shows a triage colour and a one-line summary — no file to dig through.

A queue that reads itself

Sonia starts her morning with a queue of referral orders — some handwritten, some photos taken in a hurry.

Latest updates

Who this is for

Three ways into the same platform — run it, extend it, or fund the next deployments.

Health organizations

Get back the hours your staff spend on the screen.

Booking, reminders, coverage checks and documentation stop being manual work. No licence fee, no vendor lock — you host it, you own the data.

  • Self-hosted, your data stays yours
  • Running today at the Pinamar installation
Live demo Contact
Developers

Ship a whole country as a module, not a fork.

Identity, coverage and health-record connectors are pluggable, so your region’s integration is a pull request instead of a rewrite.

  • Pluggable per-country connectors
  • A codebase already carrying real clinical traffic
GitHub
Investors and grant organizations

Every dollar buys deployments, not licences.

Funding goes to hardening, independent clinical validation and the next public pilots. Because the platform is free to run, the same grant reaches more clinics.

  • No licence cost between grant and clinic
  • Outcomes stay open for anyone to check
Contact