ChatGPT Prompts for Speech Therapists: Why Generic AI Fails SLPs (And What Actually Works)

Generic ChatGPT will write a SOAP note for a speech therapist — but the output reads like it was written by someone who has never set foot in a therapy room. Specialized prompts built around SLP-specific terminology, goal structures, and documentation standards produce notes you can actually sign off on.

---

The Problem: Generic Prompts Put the Burden Back on You

You open ChatGPT after a back-to-back day of sessions. You type: "Write a SOAP note for a speech therapy session."

What comes back is a generic paragraph that uses "client" but misses the objective data you collected, skips the ICD-10 context, lumps subjective observations together with measurable trial data, and uses language that would raise eyebrows in a peer review. You spend the next ten minutes editing it into something usable — at which point you might as well have written it yourself.

This is the core failure of generic AI for SLPs: the tool does not know what a 504 accommodation looks like versus an IEP goal, does not know the difference between an articulation target and a phonological process, and has no concept of the +/- accuracy scaffolding language that your facility expects. The result is not a shortcut — it is a first draft that still requires expert knowledge to rescue.

Specialized prompts flip this. They front-load the SLP context so the AI produces output that is 80–90% ready to use instead of 40%.

---

What Generic ChatGPT Actually Produces (Before)

Here is a realistic exchange using a bare, generic prompt — the kind most SLPs try first.

Generic prompt:

"Write a SOAP note for a speech therapy session with a 7-year-old working on articulation."

Typical generic output:

S: Client participated in speech therapy session and was cooperative. O: Client worked on articulation skills with therapist guidance. A: Client is making progress toward articulation goals. P: Continue therapy as recommended.

This note would fail audit at most facilities. It contains no objective trial data, no error pattern description, no specific phoneme targets, no cueing hierarchy used, no measurable baseline or response, and no plan update tied to performance.

---

What a Specialized Prompt Produces (After)

Specialized prompt (copy-paste ready):

"Write a SOAP note for a pediatric speech therapy session. Client: 7-year-old male, dx phonological disorder, target: final consonant deletion. Session data: 18/25 correct (72%) in structured drill with minimal verbal cue. Previous session: 14/25 (56%). Cueing hierarchy used: model, then delayed imitation. Setting: outpatient clinic. Format: standard SOAP with objective data in O section, include accuracy percentage and cue level. Keep A section to clinical interpretation only. P section: next session goal adjustment based on 72% accuracy."

Resulting output (representative):

S: Client was alert and cooperative throughout the session. He required brief redirection x2 during drill activities but returned to task quickly. O: Client produced final consonant deletion targets (CVC words) with 72% accuracy (18/25 trials) with minimal verbal cuing (delayed imitation model). This represents a 16-percentage-point increase from the previous session (56%, 14/25). A: Client demonstrates continued progress toward phonological accuracy targets. Performance at 72% with minimal cuing suggests readiness to reduce cue support in structured contexts. P: Advance target to minimal cuing in structured phrase-level context next session. Introduce semi-structured play probe if 80% accuracy is achieved under current conditions.

That note goes directly into the EMR. The difference is not AI quality — it is prompt quality.

---

5 Copy-Paste Prompts for SLP Documentation You Can Use Today

These prompts are built to work in standard ChatGPT (free or paid). Remember: never include real client names, dates of birth, or other PHI. Use initials, age, and diagnosis only.

Prompt 1 — SOAP Note from Session Data

"Write a SOAP note for a speech therapy session. Client: [age, sex], diagnosis: [dx]. Target: [specific goal, e.g., /r/ in initial position at word level]. Trial data: [X/Y correct] with [cue level, e.g., phonemic cue]. Setting: [school/outpatient/teletherapy]. Prior session accuracy: [X%]. Format: standard SOAP. O section must include trial count, accuracy percentage, and cue level. A section: clinical interpretation only. P section: next-session plan tied to today's performance."

Prompt 2 — Progress Note for Insurance/Billing Justification

"Write a progress note suitable for insurance authorization renewal. Client: [age, dx]. Current goal: [goal text from IEP/treatment plan]. Baseline at start of authorization period: [X%]. Current performance: [X%] across [N] sessions. Medical necessity rationale: client has not yet achieved functional independence in [skill]. Keep language objective and measurable. Do not include PHI."

Prompt 3 — Parent/Caregiver Home Program Explanation

"Write a plain-language explanation of a home practice activity for the parent of a [age]-year-old working on [target skill]. Describe the activity in 3 steps, explain what the parent should look for, and give one example of how to respond when the child gets it right and one example of how to prompt when they make an error. Avoid clinical jargon. Tone: warm, practical, encouraging."

Prompt 4 — IEP Goal Language

"Write 2 measurable IEP goal options for a [age]-year-old with [diagnosis]. Skill area: [e.g., receptive vocabulary, narrative language, /s/ cluster reduction]. Format: 'Given [condition], [student name/student] will [behavior] with [accuracy]% accuracy across [N] consecutive sessions as measured by [data method].' Provide one goal at an emerging level and one at an independent level."

Prompt 5 — Discharge Summary

"Write a discharge summary for a speech therapy client. Client: [age, dx]. Duration of services: [N] months/years. Initial presenting concerns: [brief description]. Goals addressed: [list 2–3]. Final performance data: [X% accuracy on each goal]. Reason for discharge: [goal met / family request / transition to school services / etc.]. Include a recommendation section for continued monitoring or follow-up. Keep under 300 words. No PHI."

---

Why SLP Documentation Is a Uniquely Hard Case for Generic AI

Generic AI tools are trained on broad text. They understand that SOAP stands for Subjective, Objective, Assessment, Plan — but they do not understand:

• Cueing hierarchy language — the difference between "independent," "minimal cue," "phonemic cue," and "model" carries clinical and billing weight that generic AI collapses into vague phrases like "with some support."
• SLP-specific diagnoses and codes — writing about "articulation disorder" when the chart says "childhood apraxia of speech" (CAS) is not a cosmetic error; it affects treatment planning and authorization.
• Measurability standards — most documentation audits and payors require trial counts and accuracy percentages in the objective section. Generic AI writes narrative where you need numbers.
• Dual audience — SLP notes are read by other clinicians, payors, parents, and sometimes legal teams. The register has to hold up across all four. Generic prompts produce one-register output.

Prompts written specifically for SLPs encode this context into the instruction itself so you do not have to re-teach the AI every session.

---

What a Specialized SLP Prompt Pack Adds Beyond These 5

The five prompts above handle the most common documentation tasks. A full toolkit extends this to the full scope of an SLP's admin load:

• Teletherapy session notes with platform-specific language
• AAC device trial documentation
• Dysphagia/swallowing evaluation summaries
• Fluency (stuttering) baseline and progress notes
• Voice disorder session notes for medical settings
• Transition planning language for school-to-adult services
• Referral letters to ENT, neuropsychology, and audiology
• Medicaid and private insurance authorization language
• Supervisor observation documentation (for CF supervisors)
• End-of-year school progress report language

The Speech Therapist AI Toolkit at PromptsForPros contains 350+ prompts organized by documentation type, setting (school, hospital, outpatient, private practice, teletherapy), and client age range, so you pull the right prompt for the right session rather than adapting a generic one each time.

Get all 350+ prompts for speech therapists — 30-day money-back guarantee, no questions asked: https://promptsfor.pro/shop?utm_source=seo&utm_medium=article&utm_campaign=speech-therapist-ai-toolkit&utm_content=vs-default

---

FAQ

Q: Can I use these ChatGPT prompts in the free version of ChatGPT? All five prompts above work in the free tier of ChatGPT. The main constraint is context length — for longer discharge summaries, a paid plan handles the full output more reliably.

Q: Is it HIPAA-compliant to use ChatGPT for speech therapy notes? The free consumer version of ChatGPT does not have a Business Associate Agreement (BAA), which means it should not be used with any protected health information (PHI). All prompts in this article and in the toolkit are written to use only de-identified data: age, diagnosis category, and session metrics. Real client names, dates of birth, and identifying details should never be entered into a public AI tool. ChatGPT Enterprise and some clinical platforms offer BAA arrangements — verify your facility's policy before use.

Q: How is a prompt pack different from just asking ChatGPT myself? The difference is in the embedded context. A well-built SLP prompt pre-loads the cueing hierarchy, required data fields, documentation format, and audience — so the AI output is clinically usable rather than a rough draft. Writing that context from scratch every session costs more time than the AI saves. A prompt pack is a reusable library; you copy, fill in session variables, and run.

Q: Will these prompts work in tools other than ChatGPT? Yes. The prompts are plain text and work in Claude, Gemini, Copilot, and most other large language model interfaces. The principles — specificity of context, explicit format instructions, measurable output requirements — apply across tools.