Sully.ai’s AI Scribe turns natural conversations into draft clinical notes by securely capturing the visit, transcribing it with medical speech recognition, understanding clinical meaning, and formatting a draft into your note template for clinician review and signature. Here’s how it works end‑to‑end.

1) Consent & capture

• Patients approve ambient recording or phone/telehealth capture according to your clinic’s policy.

• Audio is captured from the exam room, telehealth session, or phone line and associated with the patient and encounter.

2) Medical‑grade transcription

• Speech is converted to text with speaker diarization (who said what), punctuation, and timestamps.

• Common medical terms, drug names, dosages, and abbreviations are recognized to reduce corrections.

3) Clinical understanding

• The transcript is parsed to identify problems, symptoms, history, exam findings, assessments, and plans.

• Entities (medications, allergies, labs, vitals, imaging, procedures) are extracted and normalized to standard vocabularies when possible.

• Context such as negations (e.g., “no fever”), durations, and risk factors is preserved.

4) Note drafting to your format

• Content is organized into your preferred structure (e.g., SOAP, HPI/ROS/PE, Assessment & Plan) using your clinic’s templates and smart phrases.

• Key statements from the conversation are summarized into clear, concise sentences that match your documentation style.

• Where needed, the draft includes prompts for missing elements (e.g., review of systems or decision‑making details) so you can fill gaps quickly.

5) Structured outputs

• Discrete fields are prepared alongside the narrative note, such as: problems/diagnoses (with suggested ICD‑10/HCC where appropriate), orders/tasks to queue, counseling/education topics, and follow‑up timing.

• Suggested items never post automatically; they’re queued for review.

6) Routing into your workflow

• The draft note and any suggested tasks land where you already work (EHR note workspace, chart inbox, or team queue) via your configured integrations.

• Each draft includes an evidence trail (links back to transcript snippets) so you can see exactly where summaries came from.

7) Clinician review & sign‑off

• You edit as needed, accept/decline suggestions, and finalize the note. Everything remains a draft until a human approves it.

• Edits help tune future drafts to your preferences (e.g., phrasing, section order, default templates).

8) After‑visit support (optional)

• If enabled, related outputs (patient instructions, referral letters, coding suggestions, prior‑auth data packets) are generated as drafts for staff to confirm.

What this means in practice:

• Conversations stay natural; you don’t have to dictate in rigid phrases.

• Notes are concise, consistent, and mapped to your templates, reducing time spent documenting after hours.

• You remain in control: the system proposes, you review and sign.