About
What is TherapySupervisor?
A place for clinicians in training to think through their cases — the ones still on their mind, the ones that didn't get resolved in the session or in the supervisory hour. You bring real clinical material, work through what's still murky, and arrive at a clearer understanding of what's happening between you and your patient.
What you understand stays with you — into your next session, into supervision, into how you're learning to do this work.
Is it a replacement for human supervision?
No. Supervision is irreplaceable — a relationship that knows you and the case over time, the work of two people in the same room. Software can't do that, and TherapySupervisor doesn't try to. What it can do is meet you in the hours between supervisions, when the case is still on your mind and your supervisor isn't there. What you understand from working with the tool goes back into the room with your supervisor, where the work that only your supervisor can do happens.
What does it actually do?
It works with you on the case you bring. It asks questions that help you say more about what happened. It brings in context you may not have considered yet — what else might be alive in the case, what your asking is pointing toward, what's sitting next to the question you brought. It is rigorous when your formulation is vague and stays steady when you push for a quick answer. It does not tell you what to do with a patient — that judgment is, and remains, yours.
Why not just use ChatGPT?
You can, and trainees do. The difference is the discipline. A general AI is built to satisfy — it gives the interpretation you were fishing for, especially if you ask twice. This one is built for supervision: it stays at your level, lets you set things down without losing them, and doesn't reach past what you've said. That's what makes the next supervision hour worth more.
What is the theoretical framework?
Psychodynamic, with a developmental focus.
Who is it for?
Built for clinicians in training and practice — psychiatry residents and fellows, psychology interns, social work interns and trainees, mental health counseling trainees, and early-career clinicians. Also useful for experienced clinicians who want a steady space to organize their thinking across a caseload.
How do I get access?
The practice scenarios are open — create an account and begin. Working on your own cases opens after a conversation with us; case work is longitudinal, it holds your material over time, and we open it person by person rather than by default. Write to
dave@therapysupervisor.ai.
Is it a medical device?
No. TherapySupervisor is an educational tool for clinician development. It does not diagnose, treat, or provide recommendations for patient care. Nothing it offers should be acted on as clinical advice or direction — clinical decisions belong to you and, where appropriate, your supervisor.
How It Works
How does it track development over time?
Each case is held longitudinally — the tool remembers prior conversations and builds a picture of how you think about your work across sessions. Over time it can reflect back patterns in how you bring cases, what you tend to notice, where you get stuck.
Over time, we believe development of this kind is observable — that the way a clinician writes and thinks about cases changes as capacity grows, in ways an attentive reader can see. Whether that can be measured reliably, and how it relates to the frameworks programs already use — ACGME milestones in residency, their equivalents in psychology and social work training, candidacy progression in analytic institutes — is an open research question. Current pilot work is designed to help answer it, not to presume the answer.
What happens to what I type?
It stays inside one boundary. The application, the database, and the AI model all run inside a single AWS environment under a signed Business Associate Agreement. Nothing you type leaves it. The model provider does not log, retain, or review what passes through. Sessions are visible to you and to no one else.
Can my supervisor or program see my sessions?
No. Sessions are yours. Programs that adopt the tool see cohort-level signal only — how a group is developing — never what any individual said. This is not a policy that could change quietly; it is how the system is built.
Does the tool need information about my patient?
No. It works on your experience of the clinical encounter, not the patient's record. It doesn't need patient identifiers and it is built not to accept them.
How is the no-PHI rule actually enforced?
Two ways. Before you type anything, you acknowledge the de-identification standard: first name or initial and a decade of age, nothing more. Then every message passes through two screens before it reaches the AI — a pattern check for identifiers like record numbers, dates, and addresses, and a classifier that catches subtler identifying detail. If either flags a message, it is blocked, nothing is sent, and you're asked to remove the identifier.
Is it HIPAA compliant?
TherapySupervisor is not a covered entity and is not designed to receive protected health information. Users are instructed to de-identify all case material before entering it — the same standard that applies to case conferences, consultation groups, and written case write-ups. If you need a HIPAA-compliant environment for identified patient information, use your institution's EMR. TherapySupervisor is not that tool and is not trying to be.
What about emergencies?
This is the wrong tool for a crisis. If you or a patient are in immediate danger, use your program's on-call system, call or text 988 (Suicide & Crisis Lifeline), or call 911. TherapySupervisor is for reflective thinking between sessions, not for acute clinical decision-making.
Research
What research opportunities does this open up?
TherapySupervisor was built with research in mind. The data it generates may support future study of clinician development across disciplines and training contexts. The core research question is whether structured developmental work between supervision sessions can be reliably measured, accelerated, and studied — and the architecture is built to produce signal across psychiatry, psychology, social work, and analytic training, wherever clinicians develop through supervised practice.
Different settings find different angles compelling:
Residency and fellowship programs
Does structured reflection between supervision sessions accelerate developmental progression — particularly in patient care, interpersonal communication, and professionalism? Can longitudinal language analysis track development in ways that supplement traditional evaluations, including ACGME milestones in psychiatry and equivalent frameworks elsewhere?
Academic medical centers
Can developmental stage be reliably measured through clinical language over time? What does the arc from novice to competent clinician actually look like when tracked across hundreds of cases?
Psychoanalytic and psychotherapy training institutes
What are the observable markers of deepening clinical sophistication in depth-oriented work? What experienced supervisors recognize, they recognize through formation — the open question is whether any part of that recognition can be studied without flattening it.
Internship and trainee programs in psychology, social work, and counseling
The developmental questions translate across disciplines. How does supervised practice produce capacity in trainees whose programs and training structures differ from medical residency, and what shared markers might emerge?
We're open to research partnerships. If you or someone in your program has a question worth studying, we'd like to hear it.