Skip to content

Case study / Plural Health / 2026

NeoScribe

Internal AI tool that turns clinical conversations into structured notes a clinician approves rather than writes.

NeoScribe cover
Role
Lead AI engineer
Year
2026
Stack
Large language models / Next.js
AI
drafts, clinician decides
0
records written without review

Context

Clinical documentation eats clinician time, and the easy version of an AI fix makes the dangerous trade of speed for trust. NeoScribe is an internal tool at Plural Health built the other way around: the model drafts, the clinician decides.

Approach

The pipeline turns a clinical conversation into a structured draft note. That draft never becomes a record on its own. It enters a review workflow where a clinician edits and approves, so the human stays accountable for every line and nothing is published unchecked. The interface is built to make that review fast rather than to hide it.

Outcome

The product lead called it impressively intuitive, and the demo carried itself. As an internal tool the specifics stay private; more is available on request.