Inside the Generate Notes Pass: The Four Agents That Draft a Wound Care SOAP Note

Clicking Generate Notes kicks off a four-agent pipeline. Each agent has a narrow job, and each hands off structured output to the next. The whole pass typically finishes in under a minute.

1. Summarizer

Reads the dictation, intake, wound list, and Smart Clinical Overview. Produces the subjective and objective sections — history of present illness, exam findings, current measurements. This is where the AI scribe for wound care does most of its narrative work.

2. Assessment

Generates the assessment section, linking each active wound to a working diagnosis. It also proposes ICD-10 codes for the coding agent to finalize — see the coding agent deep dive for how that handoff works.

3. Planner

Builds the plan section from the orders you placed up front, plus any recommended orders (cultures, labs, referrals) triggered by the assessment. If offloading, compression, or a skin substitute is indicated, the plan cites it explicitly.

4. Validator

Cross-checks the drafted note against measurements, orders, and diagnoses. It flags contradictions — for example, a plan that continues an antibiotic without a matching infection assessment — before you see the draft.

What you get back

Three artifacts land in the encounter: a SOAP note, a procedures note (if procedures were performed), and a plain-language patient summary. All three are editable. Nothing is signed until you review.

Why the sequence matters

Running summarizer before assessment prevents the AI from anchoring on a diagnosis it hasn't justified. Running validator last catches the small inconsistencies that trigger audits and denials.