How to Write a Progress Note with AI in 2026

TL;DR. A practical walkthrough for writing payer-ready progress notes with AI — the right structure, what to never let AI invent, and the free tools that handle it. Practical walkthrough with prompt structure and the free tool.

Progress notes are the documentation that re-justifies continued care at the points payers actually scrutinize. A note that demonstrates measurable improvement, ties change to skilled intervention, and uses defensible clinical language is the difference between continued reimbursement and a denial that ends a patient's care prematurely. Writing them by hand, every progress reporting period, for every patient on a real caseload, is a brutal time tax on clinical practice.

This walkthrough is for PTs, OTs, mental health therapists, SLPs, and any clinician who writes periodic progress notes as part of their billing cycle.

What a payer-ready progress note contains

Before you can use AI well, you need to know what reviewers actually look for:

• Period covered — clear date range since the last progress note
• Treatment summary — sessions completed, interventions used, frequency
• Measurable changes — objective comparisons to baseline and last progress note
• Functional improvement — how the measurable changes translate to real-life function
• Skilled care justification — why the intervention required your clinical expertise
• Plan adjustments — what's changing for the next period and why
• Continued necessity — why ongoing care is justified

The notes that get approved on first pass are the ones that hit these elements clearly. The notes that get denied are the ones that drift into generic language without measurable evidence.

The right prompt structure

The mistake most clinicians make on first try is asking for "a progress note" without the structured comparison data. The prompt that actually works gives the AI the before-and-after numbers and the clinical context:

<task>Write a PT progress note for a 30-day reporting period.</task>

<context>
- Patient: 52F, post-op rotator cuff repair, weeks 4-8
- Sessions completed: 8 of 8 scheduled
- Interventions: progressive ROM, scapular stabilization, isometric to isotonic
  strengthening progression, manual therapy for capsular mobility

Baseline (4 weeks post-op):
- AROM: flex 95°, abd 80°
- Strength: 2/5 manual muscle test all directions
- Pain: 4/10 rest, 7/10 with movement
- Function: unable to reach overhead, dependent for dressing

Current (8 weeks post-op):
- AROM: flex 145°, abd 130°
- Strength: 3+/5 most directions
- Pain: 1/10 rest, 4/10 with end-range movement
- Function: independent with overhead reaching to mid-shelf, independent dressing
</context>

<instructions>
- Standard progress note format
- Demonstrate skilled care language
- Tie measurable changes to functional improvement
- Justify continued care for next 30-day period
- 350 words max
</instructions>

<avoid>
- Inventing measurements I didn't document
- Generic improvement language
- Stating skilled care without explaining why
</avoid>

Notice the structure: structured before-and-after data, the explicit instructions, and the non-negotiables. The AI produces a defensible progress note; you verify the clinical accuracy.

What to never let AI do

Invent measurements. "Likely improvement" is the fastest path to a denial when an auditor asks where the data came from. Always provide the actual numbers.

Generic skilled-care claims. "Patient required skilled care" is meaningless. "Patient required skilled manual therapy to address capsular adhesions limiting end-range motion" is what reviewers approve.

Overstate functional gains. If the patient can reach mid-shelf, don't let the AI write "patient achieved full functional reaching." Be specific.

Make claims your data doesn't support. AI will sometimes round up. Watch for it.

Common mistakes

Skipping the "why continued care" section. The progress note has to justify the next period of care, not just summarize the past. Make this section explicit.

Forgetting to compare to baseline. Auditors want to see initial findings, not just current findings. Always include both.

Vague intervention descriptions. "Therex" doesn't tell the reviewer what skilled care was provided. "Progressive scapular stabilization with closed-chain interventions" does.

Length that doesn't match what your facility wants. Some facilities want 200-word progress notes, others want 600. Match your facility's expectations.

The free tools that handle this for you

Several discipline-specific progress note tools on AI Career Lab are pre-configured for the conventions each discipline uses:

• PT Progress Note Generator — built for physical therapy progress note format
• OT Progress Report Generator — built for occupational therapy progress format
• SLP Progress Report Generator — built for speech-language pathology progress notes
• Dietitian Progress Note Generator — built for nutrition documentation conventions

Free with an AI Career Lab account, capped at five runs per day on the free tier.

Where AI does not belong

A few honest non-negotiables:

• Clinical assessment is yours. AI drafts the documentation; you make the clinical calls.
• Measurements come from you, not the AI. Never let it invent.
• PHI does not go in prompts. Use placeholders.
• Final responsibility is yours. Every progress note signed under your license is your responsibility and your defense in audit.

Try it on next month's reporting period

Pick one patient up for a progress reporting period. Pull the structured before-and-after data. Run it through the tool above. The progress note is the document where the AI time savings show up in dollars saved on prevented denials.

Create your free AI Career Lab account and try the clinical tools today. No credit card.