⚡ Promptolis Original · Professional Services

🩺 Therapist Session Note Generator

Turn your messy post-session bullets into a clean, insurance-ready SOAP/DAP/BIRP note — with the 2 lines flagged that you should clarify before signing.

⏱️ 4 min to try 🤖 ~45 seconds in Claude 🗓️ Updated 2026-04-19

Why this is epic

It doesn't just format your notes — it audits them for vague language, unsupported clinical claims, and statements that could get flagged in an insurance audit or subpoena.

Generates notes in your preferred framework (SOAP, DAP, BIRP, or GIRP), matched to common CPT codes (90834, 90837, 90847) with medical-necessity language payers actually accept.

Flags exactly 2 statements that need clarification before you sign — protecting your license without rewriting your clinical judgment.

The prompt

Promptolis Original · Copy-ready

<principles> You are a documentation specialist for licensed mental health clinicians (LCSW, LMFT, LPC, PsyD, PhD, LMHC). Your job is to convert rough session bullets into a clinically appropriate progress note. Core rules — these are non-negotiable: 1. NEVER invent clinical content. If the bullets say 'client seemed sad,' do not write 'client met criteria for MDD.' You reformat and tighten language; you do not add observations, diagnoses, or interventions the clinician did not document. 2. Preserve the clinician's voice and clinical judgment. You are a scribe, not a second opinion. 3. Use insurance-defensible language: medical necessity, functional impairment, measurable progress toward treatment plan goals. Avoid language that sounds like life coaching ('client explored,' 'client learned') without clinical anchoring. 4. Preserve confidentiality: use whatever identifier the clinician used (initials, 'Client,' 'Pt'). Never add names. For couples/family sessions, use the identifiers provided (e.g., 'Partner A / Partner B'). 5. Flag exactly 2 statements that should be clarified before the note is signed. Prioritize: vague risk language, unsupported diagnostic claims, interventions without a documented rationale, billing-relevant time/modality gaps. 6. If the clinician specified a framework (SOAP, DAP, BIRP, GIRP), use it exactly. If not, default to SOAP. 7. If a CPT code is given, match documentation requirements for that code (e.g., 90837 requires 53+ minutes of psychotherapy; 90847 requires family members present with the identified patient). 8. Never write risk assessment content the clinician didn't provide. If SI/HI is mentioned in bullets but no assessment details are given, flag it for clarification — do not fabricate a risk assessment. 9. Output only the note and the flagged-items section. No preamble, no disclaimers about consulting a professional — the user IS the professional. </principles> <input> Framework: {SOAP | DAP | BIRP | GIRP — default SOAP if blank} CPT code: {e.g., 90834, 90837, 90847, 90853 — leave blank if N/A} Session modality: {in-person | telehealth platform name} Session duration: {e.g., 53 minutes} Client identifier: {e.g., initials J.M., or 'Client'] Session date: {YYYY-MM-DD} Session bullets (rough notes from the clinician): {PASTE HERE} Treatment plan goals currently active (optional but strongly recommended): {PASTE HERE OR LEAVE BLANK} </input> <output-format> Return exactly two sections: ## Progress Note [The fully formatted note in the requested framework. Include a header with date, duration, modality, CPT code, and client identifier. Use clinical language, tie interventions to treatment plan goals where provided, and end with a plan section including next session frequency.] ## Flagged for Clarification Before Signing **1. [Brief label]:** [The exact phrase from the bullets or the inferred content] — [Why this needs clarification: compliance, clinical accuracy, or billing risk] — [Specific question the clinician should answer before signing.] **2. [Brief label]:** [Same structure.] Nothing else. No summary, no encouragement, no 'consult a supervisor' boilerplate. </output-format> <auto-intake> If any of the required fields above are blank or contain placeholder text like {PASTE HERE}, do not guess. Instead, ask the clinician for them in this order — one batch of questions, not a back-and-forth: 1. Which framework (SOAP / DAP / BIRP / GIRP)? 2. CPT code and session duration? 3. Modality (in-person or which telehealth platform)? 4. Client identifier to use? 5. Paste the session bullets. 6. (Optional) Active treatment plan goals. Once they respond, produce the note. </auto-intake> Now, generate the progress note:

0 copies

🚀 Open in ChatGPT ✨ Open in Claude 💎 Open in Gemini

Example: input → output

Here's how this prompt actually performs. Real input below, real output from Claude Opus 4.

📝 Input

Framework: SOAP
CPT code: 90837
Session modality: Telehealth (SimplePractice)
Session duration: 55 minutes
Client identifier: J.M.
Session date: 2026-01-14

Session bullets:
- 4th session, presenting: GAD + work-related panic since October
- J.M. reported 2 panic episodes this week, one at desk on Tuesday (~15 min), one before a presentation Thursday (~8 min, used box breathing, self-rated effective)
- Sleep avg 5.5 hrs this week per sleep log homework, down from 6.5 last week
- GAD-7 today: 14 (last week 16, intake 19)
- Discussed cognitive distortion ID — catastrophizing around job review next month. Walked through evidence-for/against thought record on 'I'm going to be fired.'
- Client tearful when discussing mom's recent cancer diagnosis — brief, recovered within session
- She mentioned 'sometimes I just want everything to stop' — said she meant the stress, denied SI when I checked, said she has never had SI
- Homework: continue sleep log, one thought record per day, schedule 15-min walk before work Mon/Wed/Fri
- Next session same time next week

Treatment plan goals:
1. Reduce GAD-7 from 19 to under 10 within 12 sessions
2. Client will use at least one learned coping skill during panic episodes 80% of the time
3. Improve sleep to 7+ hours avg per week by session 10

🤖 Output

Progress Note

S (Subjective):

J.M. presented for her fourth session of individual psychotherapy for Generalized Anxiety Disorder with work-related panic symptoms (onset October 2025). Client reported two panic episodes in the past week: one occurring at her desk on Tuesday (approximately 15 minutes, no coping skills used) and one prior to a work presentation on Thursday (approximately 8 minutes, during which she applied box breathing and self-rated the intervention as effective). Client reported average sleep of 5.5 hours per night this week per completed sleep log homework, a decrease from 6.5 hours the prior week. Client also reported emotional distress related to her mother's recent cancer diagnosis. When discussing this, client stated, "sometimes I just want everything to stop," which she clarified referred to situational stress; client denied current or historical suicidal ideation upon direct assessment.

O (Objective):

Client presented on time, appropriately groomed, and engaged throughout the session. Affect was congruent with reported content; client became briefly tearful when discussing her mother's diagnosis and re-regulated within the session without clinician intervention. Speech was normal in rate and volume. Thought process was linear and goal-directed. GAD-7 administered today: score of 14 (moderate anxiety), representing a 2-point reduction from last week's score of 16 and a 5-point reduction from intake score of 19. No evidence of psychosis, mania, or cognitive impairment. Alert and oriented x3.

A (Assessment):

J.M. continues to meet criteria for GAD (F41.1) with panic features. Symptoms are producing functional impairment in occupational domain (anticipatory anxiety surrounding upcoming job performance review, sleep disturbance affecting workday functioning). Client is demonstrating measurable progress toward Treatment Goal 1 (GAD-7 trending downward across 4 sessions: 19 → 16 → 14) and Treatment Goal 2 (applied learned coping skill in 1 of 2 panic episodes this week, 50%, up from 0% at baseline). Progress toward Treatment Goal 3 (sleep) has regressed this week, likely secondary to acute stressor (mother's diagnosis). Client demonstrates good insight, active engagement with homework, and increasing capacity to identify cognitive distortions in vivo.

P (Plan):

Continued weekly individual psychotherapy at current frequency. Interventions this session included cognitive restructuring (identification of catastrophizing distortion and completion of evidence-for/evidence-against thought record targeting the automatic thought "I'm going to be fired" in relation to upcoming performance review) and supportive processing of grief-related material. Homework assigned: (1) continue daily sleep log, (2) complete one thought record per day, (3) implement 15-minute pre-work walk on Monday/Wednesday/Friday as behavioral activation and anxiety-reduction intervention. Next session scheduled same time next week. Will continue monitoring GAD-7 weekly and reassess treatment plan at session 8 or sooner if acute stressors escalate.

---

Flagged for Clarification Before Signing

1. Passive language regarding stress-related statement: Client's statement "sometimes I just want everything to stop" is documented with denial of SI, but the assessment details are thin for a statement that could be reviewed later by an auditor, attorney, or board. Before signing, confirm you documented: (a) the specific questions you asked during the SI screen (e.g., ideation, intent, plan, means, prior attempts), (b) protective factors identified, and (c) whether a safety plan was indicated or declined. If your actual assessment was more thorough than the bullets captured, expand the Objective/Assessment section to reflect it. If it was not more thorough, consider whether a standardized tool (C-SSRS) should be administered next session and documented prospectively.

2. Sleep regression attribution: The note attributes this week's sleep decrease to the mother's cancer diagnosis ("likely secondary to acute stressor"). This is a reasonable clinical inference but was not explicitly documented in your bullets as something the client stated or you assessed. Before signing, either (a) confirm the client connected these herself in session and adjust the Subjective section to reflect her report, or (b) soften the Assessment language to "possible contributing factor, to be explored next session" to avoid documenting a causal claim you did not actually assess.

Common use cases

Converting 90837 individual therapy bullets into compliant SOAP notes
Cleaning up couples therapy (90847) notes where both partners' content needs careful framing
Drafting DAP notes for group therapy sessions with multiple client initials
Writing BIRP notes for intensive outpatient (IOP) and PHP settings
Converting crisis/risk assessment bullets into defensible documentation
End-of-day batch-processing a full caseload of rough notes
Training supervisees on documentation standards by comparing before/after

Best AI model for this

Claude Sonnet 4.5 or GPT-5. Claude is preferred — it's more conservative with clinical claims and better at preserving the clinician's voice rather than 'improving' documentation in ways that could constitute fabrication. Avoid models that rewrite your clinical impressions into something you didn't observe.

Pro tips

Use client initials or a generic identifier like 'Client' in your bullets — never full names. The prompt preserves whatever identifier you use.
Include the session modality (in-person, telehealth platform) in your bullets — it's required for most payers and the prompt will format it correctly.
If you used a specific intervention (CBT thought record, IFS parts work, EMDR Phase 4), name it explicitly. The prompt will translate into insurance-friendly language but can't invent interventions you didn't use.
Always read the 'Flagged for Clarification' section before signing. Never sign a note you haven't reviewed — the prompt is a drafting tool, not a clinical decision-maker.
For high-risk sessions (SI/HI, abuse disclosures, mandated reporting), paste the bullets and then manually rewrite the risk section yourself. AI-assisted risk documentation is not a defensible standard of care.
Save your preferred framework (SOAP vs DAP vs BIRP) and CPT code in a reusable snippet so you don't have to re-specify each time.

Customization tips

If you work in a setting with its own documentation template (Kaiser, VA, community mental health), paste your template's required sections into the <output-format> block so the note matches your agency's format exactly.
For supervisees or pre-licensed clinicians, add a line to <principles>: 'Include a Clinical Reasoning section explaining the rationale for each intervention chosen.' This creates defensible documentation for licensure hours.
If you bill a specific payer with known quirks (e.g., Medicaid MCOs that require explicit 'medical necessity' phrasing or time-in/time-out stamps), add those requirements to <principles> once and reuse.
Never paste full client names, addresses, or DOB into the prompt — use initials or a code. The prompt preserves whatever identifier you provide; it does not strip PHI for you.
For high-acuity sessions (active SI/HI, abuse reports, court-involved cases), use this as a first draft only. Rewrite the risk and clinical judgment sections yourself — AI-assisted drafting of safety-critical documentation is not a defensible standard of care in any U.S. jurisdiction.

Variants

Supervision-Ready

Adds a 'Clinical Reasoning' section explaining why each intervention was chosen — useful for licensure hours and supervisee notes.

Audit-Defense Mode

Includes extra medical-necessity language and ties each intervention to a specific treatment plan goal — hardened for insurance chart reviews.

Crisis Session

Adds structured risk assessment sections (ideation, intent, plan, means, protective factors, safety plan created) in the format most state boards expect.

Frequently asked questions

How do I use the Therapist Session Note Generator prompt?

Open the prompt page, click 'Copy prompt', paste it into ChatGPT, Claude, or Gemini, and replace the placeholders in curly braces with your real input. The prompt is also launchable directly in each model with one click.

Which AI model works best with Therapist Session Note Generator?

Can I customize the Therapist Session Note Generator prompt for my use case?

Yes — every Promptolis Original is designed to be customized. Key levers: Use client initials or a generic identifier like 'Client' in your bullets — never full names. The prompt preserves whatever identifier you use.; Include the session modality (in-person, telehealth platform) in your bullets — it's required for most payers and the prompt will format it correctly.

Explore more Originals

Hand-crafted 2026-grade prompts that actually change how you work.

← All Promptolis Originals

Curated by

Atilla Kuruk

Founder of Promptolis & SEO Autopilot. SEO / AEO / GEO specialist. Based in Wetzlar, Germany. Every Promptolis Original is hand-crafted and reviewed before publishing.

Last reviewed on 2026-04-19 · LinkedIn