We are building
medical superintelligence.
State-of-the-art clinical datasets and RL environments for research labs. An AI-native EHR, with custom models inside it, for hospitals and clinics.
Backed by Y Combinator
What we offer
Frontier AI and biomedical research
For research labs
Clinical data and training environments, sourced and built for the capability you are chasing.
Hospitals and independent clinics
For hospitals and clinics
We build the record system your clinics run on, and deploy specialist models inside it.
Who it is for
Most clinical AI is trained on a narrow slice of the world: a handful of high-income countries, and mostly patients of European descent within them.
A model fitted to that slice does not generalise off it. Accuracy falls on the patients the data left out, which makes it both less fair and, simply, less good.
The EHR
A record system built for agents.
- Prior authorisation and coding
- Requests assembled from the chart and chased, and the note, the codes and the claim generated from the encounter.
- Structured at write time
- Every clinical fact is coded and addressable as it is created, so an agent acts on the record instead of reconstructing it from a scan.
- It improves with use
- Corrections and outcomes come back to the model that made the suggestion, which is the part a bolted-on agent cannot do.
Why both sides
Care produces the data. The data produces the models. The models go back into care.
01
Care happens
Clinics run on the Osseus EHR, or connect the record system they already have.
02
The record builds
Consented imaging, notes, labs and outcomes accumulate as one structured patient timeline.
03
Models get trained
Labs license that data and our environments. We train our own specialist models on both.
04
Models go back to work
They deploy into the same clinics under supervision, and what they get right or wrong re-enters the record.
