How accurate are AI-written SOAP notes compared to clinician-written notes?

AI-written SOAP notes can achieve accuracy comparable to clinician-written notes when the system is purpose-built for mental health language and paired with clinician review. However, accuracy depends on how the tool is used and where you measure. Structure & Completeness: AI excels at consistently capturing required SOAP elements that are often rushed or missed in manual notes. Clinical Nuance: Human clinicians still outperform AI on subtle judgment, case formulation, and contextual interpretation. AI cannot infer what a client meant versus what they said. Error Profile: AI errors tend to be hallucinations (inventing details) or phrasing issues. Human errors are more often inconsistency, copy-forward mistakes, or missing compliance details due to time pressure. Best Practice: Accuracy is highest when clinicians treat AI as a first-draft generator, not a final signer. A 2-4 minute review ensures the note is both technically complete and clinically nuanced.

Can AI-written SOAP notes meet documentation standards for therapy sessions?

Yes, AI SOAP notes can fully meet professional and insurance documentation standards, provided a licensed clinician reviews and attests to the final note. Compliance Requirements: Payers and regulatory bodies require documentation of medical necessity, specific interventions, and client response. AI generates these elements consistently. The Legal Record: The note becomes a legal document upon the clinician's signature. Audit Readiness: AI-written notes that are reviewed and edited for accuracy hold up in audits just as well as fully manual notes. The key is ensuring the final version reflects what actually occurred. See how to tell if an AI SOAP note is actually clinically solid

How does Twofold generate SOAP notes from therapy sessions?

Twofold uses a multi-stage process combining ambient listening, natural language processing (NLP), and mental health-specific large language models (LLMs) to generate SOAP notes. Audio Capture: Twofold runs discreetly on a device in the therapy room, using ambient intelligence to separate client and clinician voices. Transcription: Medical-grade NLP converts speech to text, accurately recognizing diagnostic terminology. Data Extraction: The LLM identifies clinical entities, symptoms, interventions, affect, risk factors, and maps them to the appropriate SOAP categories. Structuring: Extracted data is organized into the Subjective, Objective, Assessment, and Plan framework based on clinical logic. Narrative Generation: The AI producing a complete first draft in under 30 seconds.

Can Twofold integrate with existing EHR systems for clinical documentation?

Yes. Twofold is designed to integrate seamlessly with existing EHR workflows rather than functioning as a separate application that requires copy-pasting. Workflow Integration: Twofold delivers completed SOAP note drafts directly into your EHR, populating the correct fields automatically. HIPAA-Compliant Infrastructure: All data transmission is encrypted, and Twofold signs Business Associate Agreements (BAAs) required by HIPAA. No Disruption: Clinicians use their familiar EHR interface for final review and signature.

What Happens When AI Writes SOAP Notes? Insights From 1,000+ Sessions

Therapists spend an average of 10 hours per week on documentation. That is time lost to typing, and stolen from patients, family, and rest. AI SOAP notes have emerged as the promised solution to this crisis, offering to automate the note‑taking process entirely. Explore the insights from over 1,000 therapy sessions documented by AI; on where this technology excels, where it fails clinically, and what it actually means for your practice workflow.

Why Clinical Documentation Still Creates Time Pressure for Therapists

To understand the appeal of AI SOAP notes, you must first understand the weight of the problem.

The "Pajama Time" Phenomenon

Despite EHR adoption, documentation hours haven't lessened; they've shifted to after‑hours. "Pajama Time", the unpaid work done at home, remains a primary driver of burnout.

The Mental Load: Cognitive Switching

The real drain isn't typing; it's the constant cognitive switching. During a session, therapists must simultaneously listen empathetically, formulate interventions, and mentally log data for the medical record. Every second spent thinking "I need to remember this" degrades both clinical presence and documentation quality.

Compliance and Reimbursement

Poor notes lead to denied claims and audit risks. The SOAP note is a legal record; it must be accurate, defensible, and timely. This pressure creates the anxiety that makes documentation feel like a second job.

What AI-Written SOAP Notes Mean in Modern Therapy Workflows

It is crucial to understand what AI is not. AI does not "listen" or "understand" the way a human does. Instead, it utilizes two core technologies:

Natural Language Processing (NLP): To parse sentence structure, context, and syntax.
Large Language Models (LLMs): To predict and generate text based on vast datasets of medical and conversational language.

AI essentially converts speech to data, and then data to a formatted narrative.

The Modalities

In modern practice, AI SOAP note tools generally fall into two categories:

1. Real-time AI Scribes (Ambient Listening):

How it works: These tools run on a smartphone or computer in the room. They use "ambient intelligence" to filter out the clinician's voice and the client's voice, capturing the conversation without requiring the therapist to type.
Example: A therapist places an iPad on the side table. The AI listens live, and immediately after the client leaves, a draft note appears in the EHR.

2. Post-Session Summarizers:

How it works: These tools require the therapist to record the session (or take extensive notes) and then upload the text or audio after the session for summarization.
Example: A therapist types quick bullet points during a session or uses a transcription app, then pastes that raw data into a tool that formats it into SOAP structure.

How AI Generates SOAP Notes During Therapy Sessions

Here is a technical breakdown of how conversation becomes a structured clinical document.

Stage	Technical Process
Audio Capture	Ambient voice separation algorithms filter out background noise and distinguish between the patient's voice and the therapist's voice.
Transcription	NLP converts speech to text. Advanced medical NLP models are trained on medical literature that includes diagnostic manuals and dictionaries to recognize specialized terminology.
Data Extraction	The LLM analyzes the transcript to identify “clinical entities”. It tags data points relevant to mental health: symptoms, duration, severity, interventions, risk factors, and affect.
Structuring	The extracted data is classified and routed to the correct section of the SOAP framework based on clinical logic.
Narrative Generation	The AI transitions the data from bullet points to professional clinical prose. It applies standard medical writing conventions.

Insights From 1,000+ Sessions Using AI-Written SOAP Notes

The following results from recent research reveal a clear picture of where AI delivers value.

Time and Productivity Insights From AI-Written SOAP Notes

Physicians who used AI scribes most frequently experienced the greatest time savings.
Physicians using AI scribes experienced statistically significant reductions in note-taking time outside of 7 a.m. to 7 p.m., time spent per appointment, and "pajama time" (work during personal hours).

Expert Insight: “Both doctors and patients highly value face-to-face contact during a visit, and the AI scribe supports that.” - Dr. Vincent Liu, MD, Kaiser Permanente Northern California Division of Research.

Documentation Quality Insights From AI-Written SOAP Notes

By analyzing large volumes of medical data, AI can flag inconsistencies, gaps, or possible errors in documentation. The result is more accurate and complete patient records, which support better treatment decisions and long-term health management.
When documentation takes less time, clinicians can give more attention to their patients. This creates space for meaningful conversations, thoughtful clinical decisions, and more thorough care planning.

How AI-Written SOAP Notes Change Clinician Workflow and Documentation Quality

The Shift from Scribe to Editor: The cognitive load shifts from transcription to clinical review.
Eye Contact & Rapport: Freed from staring at a screen or typing notes during sessions, Clinicians report stronger therapeutic rapport and better attunement to the client's nonverbal cues.
Recall Assistance: AI catches details that often slip away in the post-session writing. This includes specific names of relatives mentioned, exact event timelines, or verbatim quotes that support diagnostic criteria.

Limitations and Risks Observed in AI-Written SOAP Notes

While the benefits of AI SOAP notes are invaluable, it is also important to acknowledge where AI falls short.

Risk Category	Example	Mitigation Strategy
Hallucinations	AI invents a detail that was not discussed.	Read every note in full. Do not skim or assume accuracy.
Affect Misread	AI cannot see flat affect, rolled eyes, or tearfulness.	Manually add one line: "Client appeared [tearful/guarded/euthymic]."
Cultural Insensitivity	AI interprets idioms or cultural expressions literally.	Edit for cultural context and confirm meaning with the client if unclear.
Redundancy	Overuse of clinical jargon makes the note sound robotic.	Humanize the language slightly. Add a brief empathetic or narrative phrase.

Best Practices for Reviewing and Finalizing AI-Written SOAP Notes

To safely integrate AI into your workflow, follow these four review steps:

Scan for Hallucinations: Quickly scan for any event, statement, or intervention you do not remember occurring. If the AI fabricated it, delete it.
Infuse the Non-Verbal: Add one line regarding the client's appearance, eye contact, or motor activity. This is where human observation beats AI.
Verify the "P" (Plan): Ensure the plan reflects what was actually agreed upon with the client, not just a generic intervention.
Tone Check: If the note feels too concise, add a brief phrase (e.g., "Client demonstrated insight into..." or "Client appeared open to...").

How to Achieve Faster & More Accurate AI-Written SOAP Notes With Twofold

The Difference with Twofold: Unlike general-purpose AI tools, Twofold is trained specifically on mental health lexicon and clinical frameworks. It understands the difference between CBT, DBT, and psychodynamic terminology, resulting in higher first-draft accuracy.
Seamless Integration: Twofold integrates directly within your existing EHR workflow. No more switching between apps.
Customization: Users can utilise specialty-specific templates that align with their therapeutic style and note preferences. This personalization pushes first-draft accuracy, meaning less editing time and more time saved.

Conclusion

AI SOAP notes are a powerful assistive technology that allows therapists to focus on what matters most: clinical presence and quality of care. The data from 1,000+ sessions proves that AI handles the transcription, but the therapist must always remain the author.

References

Bongurala, A., Save, D., Virmani, A., & Kashyap, R. (2024, May 22). Transforming Health Care With Artificial Intelligence: Redefining Medical Documentation. Mayo Clinic Proceedings: Digital Health, 2(3).

Clark, S. (2024, October 17). Pajama Time: The Unseen Hours of Dedicated Doctors. Sara Clark Consulting.

Esquer, K. (2025, December 1). The Average Caseload for a Mental Health Therapist – How Much is Enough? The Therapist Network.

Rubio, M. (2025, April 7). Analysis: AI scribes save physicians time, improve patient interactions, and work satisfaction. Permanente Medicine.

Tierny, A., Gayre, G., Hobermann, B., Mattern, B., Ballesca, M., Wilson Hannay, S., Castilla, K., Lau, C., Kipnis, P., Liu, V., & Lee, K. (2025, March 31). Ambient Artificial Intelligence Scribes: Learnings after 1 Year and over 2.5 Million Uses. NEJM Catalyst, 6(5).

FAQ

Frequently asked questions

How accurate are AI-written SOAP notes compared to clinician-written notes?
AI‑written SOAP notes can achieve accuracy comparable to clinician‑written notes when the system is purpose‑built for mental health language and paired with clinician review. However, accuracy depends on how the tool is used and where you measure.
- Structure & Completeness: AI excels at consistently capturing required SOAP elements that are often rushed or missed in manual notes.
- Clinical Nuance: Human clinicians still outperform AI on subtle judgment, case formulation, and contextual interpretation. AI cannot infer what a client meant versus what they said.
- Error Profile: AI errors tend to be hallucinations (inventing details) or phrasing issues. Human errors are more often inconsistency, copy-forward mistakes, or missing compliance details due to time pressure.
- Best Practice: Accuracy is highest when clinicians treat AI as a first-draft generator, not a final signer. A 2-4 minute review ensures the note is both technically complete and clinically nuanced.
Can AI-written SOAP notes meet documentation standards for therapy sessions?
Yes, AI SOAP notes can fully meet professional and insurance documentation standards, provided a licensed clinician reviews and attests to the final note.
- Compliance Requirements: Payers and regulatory bodies require documentation of medical necessity, specific interventions, and client response. AI generates these elements consistently.
- The Legal Record: The note becomes a legal document upon the clinician's signature.
- Audit Readiness: AI-written notes that are reviewed and edited for accuracy hold up in audits just as well as fully manual notes. The key is ensuring the final version reflects what actually occurred.
See how to tell if an AI SOAP note is actually clinically solid
How does Twofold generate SOAP notes from therapy sessions?
Twofold uses a multi‑stage process combining ambient listening, natural language processing (NLP), and mental health‑specific large language models (LLMs) to generate SOAP notes.
- Audio Capture: Twofold runs discreetly on a device in the therapy room, using ambient intelligence to separate client and clinician voices.
- Transcription: Medical-grade NLP converts speech to text, accurately recognizing diagnostic terminology.
- Data Extraction: The LLM identifies clinical entities, symptoms, interventions, affect, risk factors, and maps them to the appropriate SOAP categories.
- Structuring: Extracted data is organized into the Subjective, Objective, Assessment, and Plan framework based on clinical logic.
- Narrative Generation: The AI producing a complete first draft in under 30 seconds.
Can Twofold integrate with existing EHR systems for clinical documentation?
Yes. Twofold is designed to integrate seamlessly with existing EHR workflows rather than functioning as a separate application that requires copy‑pasting.
- Workflow Integration: Twofold delivers completed SOAP note drafts directly into your EHR, populating the correct fields automatically.
- HIPAA-Compliant Infrastructure: All data transmission is encrypted, and Twofold signs Business Associate Agreements (BAAs) required by HIPAA.
- No Disruption: Clinicians use their familiar EHR interface for final review and signature.