Skip to main content

How does Tenor Chat work?

Understanding Tenor's AI therapy assistant

Paul Salvatore avatar
Written by Paul Salvatore
Updated over 2 months ago

Therapy often presents complex challenges where an outside perspective can be invaluable. That's why we built Tenor Chat — a 100% HIPAA-compliant AI assistant that serves as an interactive thought partner whenever you need one.

What can Tenor Chat do?

Real-time consultation

Whether you’re puzzling over a differential diagnosis, refining a treatment plan, or hunting for the right worksheet to assign, Tenor Chat is an always accessible thought partner.

  • Always on call – Ask conceptual, ethical, or modality-specific questions about any case, 24 / 7.

  • Alternative viewpoints – Receive evidence-based interpretations when you feel stuck.

  • Second set of eyes – Stress-test your case conceptualizations and interventions before sessions or supervision.

Session-Aware Insights

If you record sessions in Tenor, the assistant leverages those transcripts to provide ultra-personalized guidance tailored to both you and your client.

  • Automated session analysis – Reviews your transcripts in the background.

  • Pattern detection – Flags recurring themes, affective shifts, and therapeutic ruptures across sessions.

  • Instant quote recall – Pulls exact client language for richer analysis.

  • Tailored interventions – Suggests modality-aligned techniques matched to the client’s history, goals, and in-session dynamics.

What are different ways I can use Tenor Chat?

The options are unlimited. However, here are some common use cases we've seen from therapists using the product:

When to use Tenor Chat

What you might ask

Before a session

“How can I frame today’s agenda using ACT?”


“Remind me of the follow-up items I flagged last session.”

After a session

“Summarize the client's core beliefs based on this session.”


“Give me feedback on what I did well and what I could have improved this session.”

Documentation

“Draft a letter requesting workplace accommodations based on my client's diagnosis.”


“Pull exact quotes showing my client's progress toward their treatment goals.”

Treatment planning

“Tell me the latest research on ADHD and how it might apply to my client.”


“Draft three measurable goals that align with a CBT framework.”

Diagnostics

“Which DSM-5 criteria are still unmet for a PTSD diagnosis in this case?”


“What differential diagnoses should I rule out given the latest symptom presentation?”

Self-Development and Reflection

"How do I know if I'm using self-disclosure appropriately?"

"Give me tactical exercises to explore my countertransference with my client."

Limitations of Tenor Chat

As the therapist, you are always responsible for your client's best interests and needs. Please remember the following limitations of Tenor Chat:

  • Clinical judgment – The assistant supplies suggestions, but only you can synthesize the full client history and clinical context.

  • Non-verbal insight – Because Tenor relies on text transcripts, it may miss tone, body language, and other non-verbal cues you observed in the room.

  • Crisis response – During crises, mandated-reporting situations, or any safety concerns, follow agency protocols and seek live supervision or emergency resources.

  • Legal / supervisory guidance – Tenor Chat does not replace consultation with attorneys, ethics boards, or licensed supervisors.

How does it work?

Tenor has BAAs with the major AI providers to ensure zero-retention and zero-training for all interactions we have with their services (see How is my data handled? for more information). This makes all interactions with Tenor Chat HIPAA-compliant.

Tenor adds a layer of psychology-specific intelligence on top of these models to provide more relevant information (ex. adding information from peer-reviewed psychology papers). Tenor also adds new functionality specific to your therapeutic workflows (ex. generating a quick progress note).

Context already in the Tenor system (ex. session transcripts, modality preferences) can be automatically provided to AI models to produce more relevant and accurate output.

Did this answer your question?