Bring it to your program.

TherapySupervisor is a place for your trainees to think through cases between supervision meetings. Supplemental to your supervision model, never substitutive — and built to protect both individual privacy and the supervisor's role.

How it fits into your program
A complement to supervision, not a substitute for it

TherapySupervisor works alongside your existing supervision model. Your trainees use it between meetings to think through cases — what's still on their mind, what didn't fully resolve in the session, what they want to bring more clearly into supervision.

The supervisor relationship stays where it is: irreplaceable, central, never displaced by the tool.

The trainee's developmental data belongs to them
Each trainee has their own account. Their cases are held over time, building a picture of how their thinking develops across the year. Program directors and supervisors do not have access to individual conversations. The privacy is structural — built into how the tool works, not added as a setting that could be changed.
Flexible to your context
Programs can introduce TherapySupervisor as an optional resource, integrate it into required reflection activities, or partner with us on research. We work with each program to find what fits — your residency culture, your supervision model, your evaluation framework.
Pilot programs

We're working with programs through 2026 as design partners. Pilot programs help shape how the tool develops: we learn from how trainees actually use it and refine it based on what we see. Terms for each pilot are worked out in conversation with the program.

Research opportunities

TherapySupervisor was built with research in mind. The data it generates may support future study of clinician development. We're talking with training directors and faculty about a range of questions: whether structured reflection between supervision sessions accelerates developmental progression, whether developmental stage can be reliably measured through clinical language over time, what the observable markers of deepening clinical sophistication actually look like across hundreds of cases. Questions in this category extend naturally across psychiatry, psychology, social work, and analytic training — wherever clinicians develop through supervised practice.

If you or someone in your program has a research question worth studying, we'd like to hear it.

See full research framing →
Encoded

  1. This is software that exists outside the medical record, as its intended use is to work on the trainee's experience, not the patient's record. This also protects patient data by keeping the LLM separate.

    The tool doesn't need patient identifying information to work and is specifically built not to accept it.

    Trainees acknowledge the de-identification standard before typing — first name or initial, decade of age — and every message passes two automated screens before the data reaches any model. Flagged messages are blocked, and users are asked to reframe without the PHI.

  2. The application, database, and model inference run inside a single Amazon Web Services environment under a signed Business Associate Agreement. This is an institutional standard for working with PHI, even though the expectation is nothing will reach it. Nothing a trainee types leaves it, and none of it is used to train models.
  3. Individual sessions are visible to no one — not supervisors, not the program. Institutions receive cohort-level signal only.
  4. Every response is logged against a fixed clinical framework, so drift is detectable and correctable.
What the tool won't hold

Two of the scenarios every trainee sees on day one name the tool's hard boundaries.

Medication boundary

Medication questions — doses, interactions, what to prescribe — aren't what this tool is for. Those belong with your supervisor and your team.

Safety boundary

If something in a session left you worried about a patient's safety, that's real-time territory for your supervisor — this tool can't help you evaluate a crisis.

Interested in a pilot or want to ask more questions?

Get in touch