Oct 2026
Key lessons from our free masterclass on AI in private practice: where AI can save you hours of admin, how to use it safely and compliantly, and why it will never replace you as a therapist.
Therapists have one of the most demanding jobs in healthcare. Long sessions, heavy stories, and then hours of notes, reports and admin in the evenings and weekends. AI in private practice can take a real part of that load off your shoulders, so you have more time and energy for what matters most: your clients.
In our free masterclass, WeMind founder Sander Kamphuis walks through what AI is, where it fits in the therapeutic process, how to stay compliant, and how to get started. Professor Ad de Jongh, a leading voice in EMDR, also shares how he uses AI in his own work. At WeMind we use AI every day ourselves, and it saves our team close to two hours a day. Here are the most important takeaways for therapists.
Why AI matters for therapists now
Can you still run a private practice without AI? Of course. But the direction is clear. Clients, insurers and colleagues are increasingly using technology, and in the coming years they will expect you to as well. Therapists who start exploring AI now, and understand how to use it responsibly, will be better prepared, and better able to answer their clients' questions about it.
The good news is that many therapists have not started yet, so taking even a first step puts you ahead. And one of the clearest benefits of AI transcription in healthcare is more face to face time: when you don't have to worry about writing everything down, you can focus fully on the person in front of you.
There is also understandable fear around AI in therapy. Where does my data go? Who can see it? Is this ethical? Those questions matter, and the masterclass takes them seriously. But on one point Sander is very clear: AI is not going to replace you. You are the therapist, and you will remain the therapist. AI is there to take over the administrative burden, not the therapeutic work.
Ad de Jongh puts it the same way in the masterclass. AI is not meant as a replacement for clinical judgment, but as a tool to structure raw information and improve treatment. Human oversight remains essential, and the therapist remains ultimately responsible for interpretation, diagnosis, reporting and treatment.
What is an LLM, and why you stay in control
You don't need to be a tech expert to use AI well, but it helps to understand the basics. Tools like ChatGPT, Claude and Gemini are built on a large language model, or LLM: an AI model trained on enormous amounts of text, which predicts the most likely next word based on what you give it. That is why it can turn a raw session transcript into a structured summary or report.
There are roughly four kinds of AI tools that are relevant for therapists: general chatbots such as ChatGPT, Claude and Gemini; AI scribes that transcribe your sessions; the AI features built into many EHRs; and platforms designed specifically for therapists or EMDR clinicians. General chatbots know a lot about the world, but not much about your specific clinical work. Specialised tools are trained on better, more relevant data, and often give more useful answers.
There is one important limitation to know: LLMs can hallucinate. When an AI doesn't know the answer, it rarely says so. Instead, it makes something up that sounds convincing. Ask a general chatbot a specific question about trauma therapy, and you may get an answer that is confidently wrong. That is why the golden rule is simple: AI drafts, you decide. Never send anything an AI has written to a client or another provider without reading and approving it yourself.
Where AI fits in the therapy process
A useful first exercise from the masterclass: write down every step of your practice, from the moment a client books an intake to the moment they finish treatment. Then ask yourself, for each step, where AI could save you time, and where it should stay out.
Some steps always stay with you. You do the intake. You do the therapy, the EMDR, the sessions. AI never replaces that. What it can do is run alongside you. With an AI transcriber running during the intake, you no longer have to worry about forgetting details or writing everything down while you listen.
The real time savings come afterwards. From one session transcript, AI can draft an intake report in seconds, saving around thirty minutes compared to writing it by hand. And from that same input, it can also draft an insurance report, a SOAP note, or a short summary for your client with homework exercises. Your role shifts: instead of writing everything yourself, you review, adjust and approve.
You can go a step further. Based on an intake report you have approved, AI can suggest possible DSM-5 diagnoses and differential diagnoses for you to consider. It will never diagnose for you, but it may raise something you hadn't thought of. And if you treat many clients with similar presentations, such as PTSD, AI can learn how you usually structure a treatment plan and draft one in your style.
AI can also offer something extra to your clients. Many therapists would love to send a clear, client friendly summary after each session, but simply don't have the time. With AI, that becomes a small step instead of an extra hour of work.
Prompting tips from Ad de Jongh
The quality of what AI gives you depends largely on what you ask. Ad de Jongh calls prompting the core of working with AI: the AI does the rest, but the prompt determines the result.
His most practical tip is one many people overlook. When the output isn't what you want, don't just try again. Tell the AI what you would like to be different, and ask it to write a better prompt for you. That way you learn to prompt better along the way. In his experience, it only takes four or five sessions of giving feedback before the AI produces what you are looking for.
He also recommends feeding your AI with the knowledge you already have. Notes from workshops you attended, books you value, your own templates and examples: the more relevant information you give it, the more useful its output becomes.
Compliance: the most important part
If you take one thing away from the masterclass, it should be this. Used carelessly, AI can put you and your clients at risk. Used carefully, it is safe and enormously helpful.
The starting point is the rule we've already mentioned: AI is your personal assistant, and you are responsible for everything it produces. Let it draft, but review and approve before anything is stored or sent.
Next, make sure you meet the privacy rules where you work. In the US, that means HIPAA. In Australia, the Privacy Act. In Europe, the GDPR. When an AI tool processes personal health information on your behalf, you need a signed Business Associate Agreement (BAA) with that provider. For tools like ChatGPT and Claude, a BAA is only available on enterprise plans, so arrange that first.
Never paste personal health information into the free versions of AI tools. They are tempting because they are free, but there is no BAA, you have no control over where your data goes, and you are effectively in breach. The safest approach is to remove all names and identifying details from a transcript before it goes into any general AI system.
Finally, there is consent. As Ad de Jongh emphasises, clients need to give clear permission before a session is recorded, and you should explain what happens to the recording: where it is stored, how long it is kept, and who has access.
How it works in WeMind
There are many good tools out there, and we encourage you to explore them. In the second half of the masterclass, Sander shows how you can do all of this within WeMind, and get started in about fifteen minutes.
For an in person intake, you simply press start transcriber, and a live word for word transcript runs alongside your conversation. For remote sessions, you start a video call from WeMind, share the link with your client, and the transcriber runs on the side of the screen.
Afterwards, the AI assistant works like a compliant version of ChatGPT, trained on EMDR and therapy material provided by trainers, such as treatment plans, summaries and books. Ask it, for example, which homework fits a client who is afraid of cars, and it will ask follow up questions before giving a specific, clinically relevant answer.
With custom templates, you decide exactly what your reports look like. Want session notes of no more than 250 words, always with a homework section? Set it up once, and every report follows that format. That's especially useful in a small clinic, where you want everyone to report in the same way. You select your template, press generate, review the result, and download it as a PDF or text.
Privacy is built in. WeMind does not store your transcripts or generated documents: once you leave the session, they are deleted, and they are never used to train AI. That's why you download your report and save it in your EHR.
To make that last step easier, WeMind includes an EHR plugin. You paste the text into the plugin in your browser, and AI fills in the fields of your own EHR form based on what's in the report. You check everything and confirm. Two clicks, and around fifteen seconds per client.
And finally, the masterclass shows the AI image generator: describe a scene, and within seconds you have an image to use as exposure material or a calming resource in your EMDR session.
The takeaway
AI won't change what makes you a good therapist. What it can change is how much of your time goes to admin, and how much is left for your clients and yourself. The key is to start small, stay in control, and get compliance right from day one.
The bottom line
✅ AI supports, it doesn't replace: the intake and the therapy stay with you
✅ AI drafts, you decide: always review before anything is stored or sent
✅ One transcript, many documents: intake reports, SOAP notes, summaries and treatment plans
✅ Better prompts, better results: give feedback and let the AI improve your prompt
✅ Compliance first: sign a BAA and never put health information into free AI tools
Frequently Asked Questions
Will AI replace therapists?
No. AI can take over administrative work such as notes and reports, but the therapeutic relationship, clinical judgment and treatment remain with the therapist.
Is it safe to use ChatGPT for session notes?
Only if you are on a plan with a signed Business Associate Agreement, which for ChatGPT means an enterprise plan. Never paste personal health information into the free version.
What is an AI hallucination?
When an AI doesn't know the answer, it may make something up that sounds convincing. That's why you should always check AI output before using it.
How much time can AI save me?
It depends on your practice, but drafting an intake report from a transcript alone can save around thirty minutes per client compared to writing it by hand.
Do I need consent to transcribe a session?
Yes. Clients need to give clear permission, and you should explain where the recording is stored, how long it is kept and who has access.
Watch the free masterclass
Want to see it all in action? Answer 9 quick questions about your practice, and get instant access to the free masterclass, plus personal tips based on how AI ready you are.
