The short answer
AI progress notes are clinical notes drafted by software from structured session input — such as a clinician’s dictation, a template, or session details — that a clinician then reviews, edits, and signs. The technology, often called ambient documentation, does not diagnose, decide, or finalize anything; it produces a first draft so the clinician spends less time typing after a session and more time in it, while the signed note remains the clinician’s own clinical judgment.
What are AI progress notes?
AI progress notes are clinical progress notes drafted with the help of software rather than typed from scratch. The clinician still runs the session; the system takes some form of session-specific input — a dictated summary, structured template fields, or shorthand entered during or after the encounter — and turns it into an organized draft note in the format the program uses (SOAP, DAP, or a program-specific template). The clinician then reviews the draft against what actually happened, edits it, and signs it. Nothing reaches the chart without that review.
This category of tool is often called ambient documentation when it is built to work in the background of a session rather than requiring the clinician to stop and type. The name describes the workflow, not a different level of autonomy — the human sign-off step does not change.
How the mechanics actually work
Strip away the marketing language and the pipeline is a handful of steps:
- Session input. The clinician provides session-specific detail — dictation, notes, or structured fields — during or immediately after the encounter.
- Structuring. The system organizes that input into the note format the program uses, filling sections like presenting concerns, interventions, client response, and plan.
- Draft generation. A drafted note is produced for that specific session — not a generic template, and not reused language from another client’s record.
- Clinician review. The clinician reads the draft against their own memory of the session, corrects anything inaccurate or missing, and adjusts clinical language as needed.
- Sign-off. The clinician signs the note. Only at this point does it become part of the legal record, and the signature is the clinician’s own attestation — not the software’s.
The order matters. Skipping step 4 is the failure mode every governance policy should be written to prevent.
What it does well — and where it stops
AI documentation tools are good at a narrow, well-defined job: turning session detail into an organized draft quickly. They are not a substitute for clinical thinking.
| The tool handles | The clinician retains |
|---|---|
| Organizing input into the note format | Deciding what actually happened in session |
| Drafting narrative language | Clinical judgment, diagnosis, and risk assessment |
| Reducing typing time | Accuracy, medical necessity language, and tone |
| Consistent section structure | Final review, edits, and signature |
| Speeding up the first draft | Legal responsibility for the signed note |
Treated this way, ambient documentation is a time-shifting tool, not a decision-making one. The clinical content of the note is still the clinician’s.
Where the time savings actually shows up
The most commonly cited benefit of AI-assisted documentation is a reduction in after-hours charting — the practice of finishing notes at night or between sessions instead of during the workday. That happens because reviewing and editing a structured draft is typically faster than composing a full narrative note from a blank page. The clinician is still doing quality-control work; they are doing less first-draft composition.
This benefit is workflow-dependent. A tool that requires a clinician to stop and dictate a lengthy summary mid-session may not save time over typing directly. A tool built to work from brief, natural session input tends to preserve more of the benefit — which is why evaluating the actual input method matters more than the “AI” label itself.
Governance and PHI safeguards to check before adopting a tool
Because the input to these tools is protected health information — and in substance use programs, potentially 42 CFR Part 2 data — governance questions belong at the top of any evaluation, not the bottom:
- Where does session input go? Understand whether audio, dictation, or notes are processed and stored, and for how long, before the note reaches the chart.
- Is the data used to train shared models? Ask directly. PHI used to improve a vendor’s general-purpose model is a different risk posture than PHI processed solely to draft that one note.
- Who can access drafts before sign-off? Draft notes should carry the same access controls as signed ones.
- Does the audit trail capture both draft and edits? A defensible record shows what the system produced and what the clinician changed.
- Does it respect Part 2 boundaries? SUD counseling notes and consent-gated records need the same segregation and redisclosure controls regardless of how the note was drafted.
- Is there a clear human sign-off gate? No note should become part of the legal record without a licensed clinician’s review and signature.
A simple adoption checklist
| Step | What to confirm |
|---|---|
| 1. Input method | What session detail the tool actually needs, and how clinicians provide it |
| 2. Draft quality | Whether drafts need heavy rewriting or light editing in your program’s documentation style |
| 3. Review workflow | How sign-off is enforced in the system, not just in policy |
| 4. Data handling | Storage, retention, and model-training practices for PHI |
| 5. Compliance fit | HIPAA safeguards and, where applicable, 42 CFR Part 2 handling |
| 6. Audit trail | Whether drafts, edits, and signatures are all logged |
Where this fits inside an EHR
Native integration may reduce handoffs, but it does not by itself establish draft quality, privacy, safety, or compliance. Sunwave’s current Sunwave AI page describes SIA as native to the platform and lists clinician review before signing for supported documentation workflows. Ask the team to demonstrate input capture, draft status, editing, attribution, signatures, access, audit history, model and subprocessors, retention, training use, failure handling, and feature scope for your organization.
Frequently asked questions
Do AI progress notes replace the clinician’s clinical judgment?
No. The output is a draft. A licensed clinician reviews, edits, and signs every note before it becomes part of the record — the software does not diagnose, assess medical necessity, or finalize documentation on its own.
Is patient information safe when AI is used to draft notes?
It should be handled under the same HIPAA safeguards — and 42 CFR Part 2 protections where applicable — as any other PHI in the record, with access controls, encryption, and audit logging. Ask any vendor how session data is processed, stored, and whether it is used to train shared models.
What input does ambient documentation actually use?
Approaches vary — some tools work from a clinician’s structured dictation or shorthand, others from session notes or template fields the clinician enters. In every case the system needs some form of session-specific input; it does not generate a note from nothing.
Does AI documentation reduce after-hours charting?
That is its main practical benefit: shifting the effort from typing a full narrative note after hours to reviewing and editing a draft, which is typically faster. It does not eliminate the clinician’s review responsibility.
Sources
This article is educational and describes software capabilities and general industry practices; it is not legal, clinical, financial, or billing advice. Requirements vary by organization, payer, program, and jurisdiction. Sunwave Health is a behavioral health software platform. Schedule a demo.
