Article
A Collaborative Project Between Experts
Jon Frederickson, one of the most well-known figures in ISTDP, have been working with Stanford University collaborators to develop an AI-driven ISTDP training tool.
The tool — thtrainer.stanford.edu — is free to use and open to the public. It comes with a short 4.5-minute video tutorial explaining the core features.
How the AI Trainer Works
The trainer simulates a psychotherapeutic interaction, allowing therapists to “intervene” in real time. The AI responds based on your input, helping you see the impact of your questions, reflections, and challenges.
Unlike watching a recorded session, the AI trainer:
- Engages dynamically with you
- Reacts to your interventions as a real patient might
- Provides feedback to refine your timing, focus, and delivery
Why It Matters for ISTDP
ISTDP is an experiential, emotion-focused form of psychotherapy. Therapists must:
- Read anxiety levels and discharge pathways in real time
- Detect and clarify defenses
- Mobilize complex emotions without overwhelming the patient
- Maintain a precise therapeutic focus from the very first minute
These skills can be difficult to practice outside of live supervision. The AI trainer creates a risk-free practice arena where mistakes become learning opportunities, and interventions can be repeated until they feel natural.
What It’s Not
The developers are clear:
- This is not a substitute for therapy training with real patients and supervision
- It is not a commercial product — it’s a prototype for exploration
- It’s not a replacement for human connection, but rather a complement to it
Benefits for ISTDP Therapists
- Repetition Without Risk
You can test multiple ways of phrasing an intervention and immediately see how the AI “patient” responds. - Immediate Feedback Loop
Instant reactions help you calibrate your approach faster than reviewing a recording later. - Accessibility
Anyone with an internet connection can try it, making it an equalizer for therapists worldwide who may not have easy access to live supervision. - Skill-Specific Focus
You can deliberately work on specific skills — such as challenge, pressure, or recapping — without the complexity of a whole session.
A Glimpse Into the Future
If this is where AI is today, imagine the possibilities in 5–10 years:
- More realistic simulations based on large libraries of therapy transcripts
- Adaptive patient “profiles” that learn from your choices
- Integration with supervision platforms for collaborative review
Try It Yourself
Trainer: https://thtrainer.stanford.edu/
Tutorial Video: https://thtrainer.stanford.edu/tutorial
