Drafted ambiently, signed by the clinician
Worked daily; abnormal ones prepared for review
Assembled, evidence-linked, chased when they stall
Prepared, then approved by a clinician
Coded and filed against the verified record
We didn't survey clinicians about whether they'd want this. We've been living inside the workflow — fixing what breaks, every day, in a practice where the stakes are real.
Three things a demo can't show you.
It works on real patients, in a real regulated setting — not a sandbox.
The clinician-supervision model is how it actually runs — not a compliance slide.
The economics are real: the practice pays for the work because the work is worth more than it costs.
Retention is the number that tells you it's need, not novelty.
Phase 1 is deliberately small. We kept the cohort invite-only so every account could be watched closely and whatever broke could be fixed the same week — the opposite of chasing signups. The number we optimise for isn't how many started; it's how many are still here a month later.
The clinician funnel →clinicians in the Phase 1 cohort
using it every week
still using it four weeks in
Phase 2 opens managed accounts to clinics beyond our own practice. We'd rather show you a small cohort that stayed than a large one that didn't.
Chelmsford Health Centre is the proof point. External managed accounts are the offer we're opening — not a roster we claim to already have.
Real-EHR operation is proven on our own system. "Works on any EHR" is by design, through screen-level operation; conformance to your specific system is onboarding work, not a certification we claim.
The AI runs under clinician supervision. Autonomy on real patient records ships behind formal clinical-safety governance — clinician-supervised, never "fully autonomous."
We'd rather you check the claims than take them. Bring your clinical governance lead to the call and ask.