Inside the Six AI Agents That Draft Your Wound Care Note in Five Minutes
product
A breakdown of the six AI agents behind WoundScribe — what each one drafts, what data it reads, and where the clinician stays in control.
"AI-generated note" is vague. Here's exactly which agents do what, and where a clinician still signs off.
The six agents
| Agent | What it drafts | Inputs it reads |
|---|---|---|
| Imaging | Length, width, area, depth, tissue composition | Point-and-capture photo |
| Scribe | SOAP narrative from the visit | Ambient audio |
| Charting | Structured fields for debridement, NPWT, skin substitutes | Scribe output + templates |
| Coding | ICD-10 diagnoses and CPT procedures | Chart + note |
| Coverage | Medicare criteria check before signature | Coding + payer rules |
| Education | Patient-facing handout in plain language | Chart + diagnosis |
Where the clinician stays in control
Every agent produces a draft, not a submission. You review the note from the AI scribe for wound care, confirm the coding, and sign. The coverage advisory flags missing criteria before you sign — not after a denial.
Why six and not one
A single generalist model hallucinates on wound-specific fields — tunneling, undermining, exudate character, skin substitute application windows. Specialized agents share context but each is scoped to a task it can do reliably. That's why the drafting happens in about five minutes instead of hours, and why the output survives an audit.