How Therapists Use AI Templates to Write Progress Notes in 3 Minutes
A realistic walkthrough of how therapists in private practice use AI-assisted templates to turn raw session notes into formatted progress notes quickly, without sacrificing clinical accuracy.
The Part No One Talks About in Graduate School
You trained for years to do therapy. You know how to build a therapeutic alliance, how to recognize a rupture and repair it, how to hold space for someone going through something genuinely difficult. What nobody covered in graduate school was the hour you spend every evening staring at a blank SOAP note trying to remember what a client said in session four.
If you see six to eight clients a day, which is typical for a private practice therapist working toward financial sustainability, you are writing six to eight progress notes. Not once a week. Every single day. That is 30 to 40 progress notes per week, and at 20 to 40 minutes each, the math adds up fast.
Most therapists I talk to describe the same experience: they finish their last session, feel genuinely glad about the work they just did, and then sit down to write notes for the next 90 minutes before they can go home. It is one more thing to do before you can close the laptop. It is the part of the job that the job never ends with, and for many therapists, it is what early burnout actually looks like in practice.
This article is not about a magic shortcut. It is about a specific workflow shift that a growing number of therapists in private practice are using to get their progress notes done during session transition time rather than at the end of the day. I will walk through exactly how it works, what it looks like in practice, and where to be careful.
What Changed (and What Did Not)
The technology that makes this possible is not new. What changed is the approach to using it.
Most AI documentation tools work by taking a recording or a brief summary and generating a full note from scratch. That can save time, but it introduces a meaningful risk: the AI is making decisions about what to include, how to phrase findings, and what clinical language to use. If your notes do not explicitly mention that a client denied suicidal ideation, a generative AI might not include it, or worse, might include something you never said.
The workflow I am describing works differently. You provide the raw material: your session notes, written in whatever shorthand or full sentences you use. The template defines the structure. The AI's job is to map your words to the right sections and fill the placeholders from what you gave it. Nothing more.
It is a meaningful distinction. Your clinical observations stay clinical observations. The AI is not inventing content you never wrote. Think of it less like a ghostwriter and more like a very fast copy-editor who knows your template by heart.
Before: What a Manual SOAP Note Actually Costs
Let me make this concrete. Say you just finished a 50-minute session with a client you have been seeing for three months. She came in presenting with generalized anxiety, and today she disclosed that she had a conflict with her mother over the weekend that triggered a significant anxiety spike. You used some cognitive restructuring in session and ended with a brief breathing exercise.
Here is what writing the SOAP note manually typically looks like:
You open your EHR or a blank document. You try to remember the exact details from the session: what she said, what you said, how she responded, what the plan was for next week. You write a few sentences, realize you forgot the GAD-7 score from last session that you wanted to reference, go look it up, come back, lose the thread, start again. You spend 20 minutes producing something that meets your documentation requirements but does not quite sound like the session you actually had.
Now multiply that by six sessions.
The problem is not intelligence or clinical skill. It is cognitive load after a full day of emotionally demanding work. The brain that held space for six people all day is not the same brain that writes clean structured notes at 7 PM.
After: The Template-First Workflow in Practice
Here is what the same scenario looks like with a template-first approach.
During the session, or immediately after, you jot down your raw notes. These are not formal. They do not have to be complete sentences. They just have to capture what actually happened.
For the session above, your raw notes might look something like this:
Client reported weekend conflict with mother. Triggered anxiety spike, rated 8/10. Discussed cognitive distortions around conflict, particularly catastrophizing. Client identified the thought "she'll never understand me" as automatic. We challenged this with behavioral evidence. Breathing exercise at end of session. Client responsive, rated anxiety 4/10 after exercise. Plan: continue CBT for anxiety, focus next session on core belief work. Client agreed to practice thought records this week.
That is 70 words written while it is still fresh. You would write some version of this anyway. What changes is what happens next.
You paste those notes into your documentation tool, select your SOAP template, and the AI maps your content to the right sections:
Subjective: Client reported a conflict with her mother over the weekend that triggered an anxiety spike she rated at 8 out of 10.
Objective: Client presented as engaged and responsive throughout the session. She identified the automatic thought "she'll never understand me" and demonstrated insight into catastrophizing patterns during the session.
Assessment: Client continues to demonstrate progress with cognitive restructuring techniques. Anxiety ratings decreased from 8/10 to 4/10 following breathing exercise. Core belief patterns around interpersonal acceptance remain an active treatment focus.
Plan: Continue CBT for generalized anxiety. Focus next session on core belief work. Client agreed to complete thought records as between-session practice.
That note took under three minutes from raw input to formatted output. You review it, confirm it reflects what actually happened, make any adjustments, and export it to PDF or copy it to your EHR.
The note sounds like you because it is based on your words. The structure is correct because the template enforces it. Nothing was invented because the AI only had your 70-word input to work from.
SOAP, DAP, BIRP: How Templates Handle Different Formats
One of the practical advantages of a template-first approach is that the format follows the template, not the tool's preferences. Different practitioners, supervisors, and insurance payers require different note structures. The most common for therapists in private practice are SOAP, DAP, and BIRP.
SOAP (Subjective, Objective, Assessment, Plan) is probably the most familiar. It is common in medical and mental health settings and maps naturally to the flow of a session.
DAP (Data, Assessment, Plan) is common in community mental health and counseling settings. The Data section combines both subjective and objective content, which some therapists find more natural for session documentation.
BIRP (Behavior, Intervention, Response, Plan) is structured around what you observed, what you did, how the client responded, and what comes next. It is particularly common in settings focused on behavioral outcomes or with supervisors who want intervention-level specificity.
Each of these formats has a different structure, but they are all built on the same raw material: your session observations. A good template system lets you keep one set of notes and generate different format outputs depending on who is receiving the documentation.
For a therapist who documents in DAP for her private practice but needs a BIRP format when she supervises interns, this means the same session notes can populate two different templates without rewriting anything.
The Question Every Careful Therapist Asks
At this point, most therapists I talk to ask some version of the same question: "How do I know the AI didn't make something up?"
It is the right question to ask. Hallucination risk is real in AI documentation, and the consequences in a clinical record are not minor. A note that includes something you never observed, never said, or never planned is a liability problem and a potential clinical harm.
The answer depends entirely on how the tool is designed. In a generative AI system that writes a full note from a short prompt, the risk is higher because the model has to fill gaps with plausible-sounding content. In a template-first system, the risk is structurally lower because the AI is constrained by what you provided.
The practical test is simple: give the tool an incomplete note and see what happens. If a required section of your template has no corresponding input from you, a well-designed template-first tool will leave that section blank or flag it. It will not fill the gap with a plausible guess.
With NotuDocs specifically, the template defines the placeholders, and the AI only fills a placeholder if it finds matching content in your notes. If the Assessment section of your SOAP template requires a clinical impression and you did not write one, the tool flags it rather than inventing one. You are always the clinician making the clinical call.
The second part of the answer is the review step. No responsible documentation workflow skips it. The note the AI generates is a draft. You read it before you sign it. In practice, this review takes about two minutes because you are not reading from scratch, you are confirming that the structured version matches the session you remember. The review is fast and high-quality because you are checking rather than generating.
A Realistic Day: What the Workflow Looks Like at Scale
Let me walk through what this looks like across a full day of six sessions.
Between sessions, you have a 10-minute break. You open your notes app, type 50 to 80 words capturing what happened. You do not structure them. You do not format them. You just capture the session while it is still in your head.
At the end of the day, before you leave the office, you spend 20 minutes on documentation. You paste each set of raw notes into your documentation tool, select the template, review and confirm the output, and export or copy to your EHR. Six notes in 20 minutes.
Compare that to the same therapist spending 20 to 30 minutes on each note individually at the end of the day. The difference is not marginal. It is the difference between staying an extra two hours and leaving on time.
The reason this works is that the cognitive-heavy part, observing, interpreting, and forming clinical impressions, happens during the session when you are already doing it. The raw note captures it immediately after. The template does the formatting work. You are not doing clinical thinking and formatting work simultaneously at 7 PM. You are just confirming at the end.
Practical Next Steps
If you want to try this workflow, the place to start is your notes. Before you evaluate any tool, spend one week writing raw post-session notes in a notes app or even a text file. Just capture what happened, 50 to 100 words, immediately after each session. Do not structure them. Notice how much faster that capture is when it is fresh.
Once you have that habit, the tool becomes much easier to evaluate. You are looking for a tool that takes your raw notes as input, applies your template, and does not add anything you did not provide.
NotuDocs offers a free tier with three templates and three notes per month, which is enough to run this workflow through a real week of sessions before committing to anything. The paid plan is $25 per month per seat, designed for independent practice rather than enterprise teams.
The goal is not to change the way you do therapy. It is to change how long it takes to document it, so that the work ends when the session ends.
Related reading:
