Present — AI-Assisted A&P Notes

Present AI-Assisted A&P Notes

Ephemeral

Loading model…

CONFIG

Settings

Model Selection

Choose models based on your hardware. Larger models are more accurate but require more memory and take longer to download. Changes take effect after clicking Reload Models.

Language Model (LLM) Used to clean transcripts and generate structured notes Whisper Speech-to-Text Model Larger models transcribe medical terminology more accurately

Calculator Mode

Clinical text stays in browser memory, model inference runs locally, and the service worker blocks third-party network requests after models finish loading. First-time model downloads still require the remote model hosts listed in this app's security policy.

Ephemeral clinical session Clear transcripts, audio chunks, and generated notes from page memory on clear, reload, or close. De-identify before generation Replace common identifiers such as dates, phone numbers, emails, addresses, MRNs, and SSNs before the local LLM pass. Auto-copy generated notes When off, notes only enter the system clipboard after an explicit copy click.

Medical Terminology

One term per line. During transcript cleanup, the LLM will use this list to correct any word that appears phonetically garbled or oddly spelled in a medical context — medications, diagnoses, procedures, anatomical terms, brand names, or anything else that Whisper commonly mishears. Add or remove terms based on what your practice uses. No need to predict how a word was misheard — the LLM fuzzy-matches against this list automatically.

Term list One term per line — applied during Pass 1 transcript cleanup

System Prompts

These prompts are sent to the LLM on every request. The medical terminology list above is appended to the cleanup prompt automatically — no need to add it here.

Pass 1 — Transcript Cleanup Prompt Cleans raw ASR before structuring. Medical terminology list is appended automatically. Pass 2 — Note Generation Prompt Structures the cleaned dictation into A&P notes. The boilerplate trigger list below is appended automatically — do not duplicate it here.

Boilerplate Entries

Each entry has a key (used in [BOILERPLATE:KEY] tags), optional trigger phrases that tell the LLM when to emit the tag, and the boilerplate text injected into the output.

INPUT

Assessment & Plan

Click to begin recording

Transcript

OUTPUT

Structured Notes

Structured notes will appear here after processing