Back to Blog

Blog

Therapy 1.0, 2.0 and 3.0: Where AI Between Sessions Fits

Therapy 1.0, 2.0 and 3.0: how therapy delivery has changed, where AI support between sessions fits, and what the research has tested so far.

January 16, 2026
5 min read
By Declan Ahern
AI in therapyfuture of therapyclinical practicebetween-session supportTherapy 3.0

AI support between sessions is the third change in how therapy reaches a client, after the weekly session in a room and the move to video and written contact. This post describes the three stages, labeled Therapy 1.0, 2.0 and 3.0, and what the research has tested so far.

In 2020, at the peak, 64% of US psychologists responding to the American Psychological Association's annual practitioner survey offered fully remote practice. By 2023 that share was 21%, and 67% worked in hybrid practices, seeing some clients in person and others remotely.1 Seeing a therapist on video is now an ordinary part of care.

Video changed where a session takes place. AI allows a different change: software that works with a client in the days between sessions, under their therapist's direction. The three labels group a few changes in how care reaches a client and leave out most of the field's history.

Three stages of delivery

Therapy 1.0 is the weekly session in a room. Between appointments the client has homework, memory and whatever support they find on their own, and the therapist learns how the week went when the client tells them.

Therapy 2.0 puts therapy on screens. Sessions move to video, and contact can move into writing. A 2013 review of 123 studies of online counseling discusses the time delay between written messages, and the way people write online, as features of the work. It concludes that online counseling can have a similar impact to face-to-face work, and that stronger evidence is still needed.2 Stage 2.0 also includes self-help programs and apps that a client works through alone. So in 2.0, help between sessions either takes the therapist's own time, as a written reply does, or comes from a program the client uses by themselves.

Therapy 3.0 is software that works with the client between sessions, set up and supervised by the therapist. It differs from 2.0 in two ways. The software does some of the between-session work itself, so that support does not have to come out of the therapist's hours. And the therapist decides what it does and can see what happened, so the week reaches the next session through more than the client's memory of it.

StageWhere sessions happenHelp between sessionsWhat the therapist learns about the week
Therapy 1.0In a room, weeklyHomework and the client's own resourcesWhat the client says in the next session
Therapy 2.0On video, or in writingA written reply from the therapist, or a program the client uses aloneWhat the client says, plus any messages exchanged
Therapy 3.0In a room or on videoSoftware the therapist sets up and directsWhat the client says, plus what the software recorded, when they choose to look

What supervised means in practice

In Citt for therapists, the therapist sets up the AI support for each client and chooses its approach, such as CBT, DBT skills or ACT, or writes their own instructions; choosing the approach the AI support follows covers how. The therapist can also send assessments and worksheets between sessions, described in what a therapist can set up between sessions. They can read a client's conversations with the AI when they choose, and before each session Citt summarizes check-in trends, conversation themes and assessment changes since the last one. Sessions can be recorded and transcribed into a draft note, and no note reaches the chart until the therapist approves it; how AI drafts progress notes goes through that step.

The therapist is not expected to watch chats in real time or to answer outside practice hours. Client messages are screened for risk before the AI replies, and how that screening works is covered separately.

What the research has tested

An early real-world study of the Wysa app, published in 2018, compared 108 frequent users with 21 infrequent users who reported depression symptoms and found that the frequent users improved more.3 It had no control group, and all three authors had ties to the company. In 2025 a randomized trial of Therabot, a generative AI chatbot, assigned 210 adults with depression, anxiety or a high risk of eating disorders to four weeks of the chatbot or to a waitlist. The chatbot group had greater symptom reductions in all three groups, at four weeks and at eight.4

Both are studies of stage 2.0 tools: a program the client uses alone. If a chatbot on its own reduced symptoms more than a waitlist, a reader can fairly ask what the therapist adds. The Therabot trial compared the chatbot with a waitlist. It did not compare it with therapy, or test it as part of a therapist's care.

I think AI between sessions belongs inside a therapist's care because someone accountable for the client can see what it did, change its instructions, and take up in session what came up during the week. Neither study tested AI support alongside a therapist, and Citt has no evidence yet that it improves clinical outcomes. The question stage 3.0 has to answer is whether clients of a therapist who uses this kind of software do better than clients of one who does not.

Footnotes

  1. The remote and hybrid figures are from APA's 2023 Practitioner Pulse Survey, reported in Mental health crisis highlights access challenges (American Psychological Association press release, December 5, 2023). 561 licensed US psychologists responded, from a random sample of 16,557 emailed in August and September 2023. ↩

  2. Richards, D., & Viganó, N. (2013). Online counseling: A narrative and critical review of the literature. Journal of Clinical Psychology, 69(9), 994-1011. A narrative review of 123 studies. The abstract lists time delay and writing behavior among the features of online counseling it discusses. The authors describe the evidence as positive and say stronger empirical evidence is still needed. ↩

  3. Inkster, B., Sarda, S., & Subramanian, V. (2018). An empathy-driven, conversational artificial intelligence agent (Wysa) for digital mental well-being: Real-world data evaluation mixed-methods study. JMIR mHealth and uHealth, 6(11), e12106. An observational study of anonymous app users comparing high and low users on the PHQ-9, without a control group. One author is a Wysa employee and one a paid consultant; the first author is a scientific advisor. ↩

  4. Heinz, M. V., Mackin, D. M., Trudeau, B. M., et al. (2025). Randomized trial of a generative AI chatbot for mental health treatment. NEJM AI, 2(4). 210 adults randomized to a four-week intervention or a waitlist, with follow-up at eight weeks. The waitlist group had no app access during the study, and the trial did not include a therapy or combined-care arm. ↩

Frequently Asked Questions

What do Therapy 1.0, 2.0 and 3.0 mean?

Therapy 1.0 is the weekly session in a room. Therapy 2.0 moves sessions and written counseling onto screens and adds self-help programs used alone. Therapy 3.0 is software that works with the client between sessions, set up and supervised by the therapist, who can see what it did.

Is there evidence that AI co-pilots improve therapy outcomes?

The studies cited here tested AI chatbots used on their own: an observational study of the Wysa app and a 2025 randomized trial of Therabot against a waitlist. Neither tested AI support alongside a therapist, and Citt has no outcome evidence of its own.

Does a clinician review every conversation a client has with the AI?

No. The therapist can read a client's conversations when they choose and is not expected to watch them in real time. Every message is screened for risk before the AI replies.

Does Citt replace the therapist?

No. The therapist chooses what the AI support does for each client, approves every note before it reaches the chart, and stays responsible for clinical care.

Stop rewriting notes from scratch

Citt.ai drafts session notes under your review, keeps between-session context in one place, and stays therapist-led. Start free. No card required.