The consult flow
How a consultation moves from recording to exported note.
Every consultation follows the same pipeline. This page describes each stage and what you control at each one.
Pipeline overview
| Stage | What happens | Your role |
|---|---|---|
| Record | Audio is captured and transcribed live (browser-first, server fallback) | Speak naturally; confirm consent first |
| Structure | A structured anamnesis draft is generated from the transcript | Review every field |
| Review | You edit transcript and anamnesis until the note is accurate | Approve or keep editing |
| Export | The finalized note is rendered to PDF | Download and file it |
Recording
KTB supports several capture sources — in-browser recording, meeting-bot capture, dictation, and audio upload — all feeding the same transcript and anamnesis pipeline. Pick whichever fits the encounter.
Recording a patient encounter requires their consent. Declare it in the app before you start — the timestamp is kept as part of the compliance trail.
Review and approval
The anamnesis is a draft produced by machine extraction. Treat it like a scribbled first pass: correct names, doses, dates, and anything the model paraphrased. Nothing becomes part of the official record until you approve it.
When the transcript was re-processed server-side, the detail page shows both the raw transcription and the cleaned version side by side so you can compare.
Export
Finalized consultations export to PDF. The export contains what you approved — it is a documentation artifact, not a clinical recommendation.
If something stalls
Audio processing is asynchronous: after recording, the audio moves through stored → queued → processing → done (or failed). A badge or status on the consultation shows where it is; failures surface an error reason without ever exposing the audio itself. See Troubleshooting.