PREECURSOR
Healthcare & life sciences

Clinical documentation that gives clinicians back their evenings

An illustration of ambient, grounded note generation designed to return time to clinicians without compromising coding accuracy or the chart.

The challenge

What stood in the way

Clinicians across the network were losing two-plus hours every day to documentation — much of it after their last patient, the so-called pajama time that drives burnout and attrition. Earlier generative pilots had been pulled because they hallucinated findings and miscoded encounters, exactly the failure modes a clinical setting cannot tolerate. Any system had to be measurably accurate against the chart and defensible under audit before it could touch a patient note.

What we build
01
Ambient note generation
The system drafts structured notes from the visit and the existing record, keeping terminology, coding, and the chart accurate rather than fluent-but-wrong.
02
Provenance on every claim
Each statement in a note links back to its source in the encounter or record, so clinicians verify in seconds and quality teams can audit after the fact.
03
Clinical eval harness
We built an eval suite tuned to clinical risk — citation accuracy, coding correctness, omission and hallucination rates — and gated every release on it.
04
Specialty-aware rollout
We tuned and rolled out specialty by specialty with the clinicians in the loop, so adoption stuck instead of stalling at the first bad note.
What it's designed to do

Notes are drafted from the visit and the existing record, keeping terminology, coding, and the chart accurate rather than fluent-but-wrong.

Every statement links back to its source in the encounter or record, so clinicians verify in seconds and quality teams can audit after the fact.

A clinical eval suite gates every release on citation accuracy, coding correctness, and hallucination rate — so accuracy is measured, not asserted.

Could we build something like this for you?

This is an illustrative example of the kind of system we build. Tell us the metric you need to move and we will tell you, candidly, what we would build first.

See more work