The AI+ CBT Lab Program

Mental wellbeing at scale.

Learn below about our new AI-enhanced group mental health support program that could surpass the most cost-effective interventions currently available. Then, discover how you can help us launch it.

The problem

A vast global mental health gap

No sufficiently accessible, cost-effective, and scalable solutions

  • In-person interventions such as StrongMinds (promoted by the Happier Lives Institute as one of the most impactful mental health programs) demonstrate results, but their growth is constrained by their reliance on local human supporters.

  • Fully online services like chatbots and self-help apps lack the human connection needed to maintain engagement. They reach millions, but 50-90% of users abandon them within days  (Torous et al., 2020).

Make it happen.

Donate

We need $100,000 by the end of November 2025.

Every dollar counts.

Donate now

Advise us

We are actively recruiting.

  • Technical experts, specifically in AI agent development and data security

  • LMIC locals with connections to higher education, healthcare, or policy organizations

We would love to hear from you!

Get in touch

Our solution

AI-enhanced online group mental health support

The AI+ CBT Lab is a 12-week digital mental health program where groups of 4-6 people meet weekly with a trained facilitator to practice evidence-based psychotherapy and behavior change techniques—while AI handles ~80% of behind-the-scenes operations through LLMs and automations. 

Participants get

  • Weekly video sessions with their group and facilitator

  • Evidence-based workbook with Third-Wave CBT techniques

  • Peer community for support between sessions

  • 24/7 AI coach for questions, exercise reviews, and personalized encouragement

AI handles

  • Recruiting and matching compatible groups

  • Training facilitators through personalized coaching

  • Scheduling, reminders, and routine questions

  • Progress monitoring and risk flagging

Human experts focus only on

  • High risk or complex cases

  • Program design

  • Clinical oversight

We aim to deliver the best of both with the limitations of neither

Service

Strong Minds

Woebot

RW CBT Lab program

RW AI+ CBT Lab program (predicted outcomes)

Human Component

  • Peer-group

  • Facilitator

  • Facilitator trainer (practitioner)

None

  • Peer-group

  • Facilitator

  • Customer support

  • Facilitator trainer

  • Moderator

  • Online community

  • Peer-group

  • Facilitator

  • Online community

Digitisation

  • Home practice app

100%

  • Workbook

  • Monitoring

  • Habit & mood tracking apps

  • Workbook

  • Monitoring

  • Habit & mood tracking apps

  • AI moderator

  • AI participant support

  • AI monitoring

Scalability

Moderate (due to in-person) limitations

High

Moderate- high

Likely high

Costs

Low

Very low

Moderate

Likely very low

Engagement

High

Very low

High

Likely high 

Effectiveness

High, equivalent to 1:1 therapy

Moderate, for those who finish

High, equivalent to 1:1 therapy

Likely high, equivalent to 1:1 therapy

Meet Amara: A sample participant

Amara, 24, Lagos  

(Example participant, for reference only)

Week 0

Amara was feeling overwhelmed by constant worry about her job performance and financial future. The anxiety kept her up at night and made it hard to focus during the day. She felt stuck, unable to take on new opportunities despite wanting to grow her career. She learns of the AI+ CBT Lab and decides to apply. She fills out an application form online and completes a brief mental health screening. AI matches her with 5 others of similar age, career stage, availability, and goals.

Week 1

Amara has her first session via video call. Her facilitator, tested and trained by a dedicated AI agent, guides introductions and teaches the CBT cycle. Between sessions, Amara uses the workbook and asks the AI coach questions when she's stuck. The AI sends her reminders to practice her exercises.

Weeks 2-8

Each week, Amara builds new skills—challenging negative thoughts, practicing mindfulness, and building healthy habits. She posts her home practice reviews weekly to the online community, where her group cheers her on. One week she almost forgets, but her accountability buddy sends a gentle reminder just in time! When Amara struggles with an exercise, the AI coach walks her through it. When she shares a breakthrough in the community, her group celebrates with encouragement and their own insights.

Week 12

During the final reunion session, Amara has completed 95% of exercises and reports significant improvements in anxiety and productivity. She's gained 6 hours of productive time per week—enough to take on a side project that supplements her income and moves her toward her career goals.

Total program cost: $18-80
Outcomes: ~2 WELLBYs over 4 years, reduced suffering, and increased work productivity

Image generated with ChatGPT5.0

How? A Roadmap

Phase 1 | Months 0-8 | Building and Testing

$100,000 funding needed

  • Design AI tools, safety protocols, and estimate costs

  • Build AI facilitator coach, and a participant coach

  • Create matching, scheduling, and monitoring systems

  • Test with the existing community, including LMIC participants

  • Validate 85% retention and cost targets

  • Finalize rollout plan, build partnerships, fundraise, and prepare publication

Phase 2 | Months 9-16 | Entering the Market

 $116,000 additional funding needed

  • Recruit LMIC experts with lived experience and build partnerships

  • Culturally adapt program for specific communities and countries

  • Run early adopter pilot with English-speaking digital natives

  • Refine and begin scaled rollout

Why us?

Proven results

Our flagship CBT Lab has been tested and replicated across 150+ participants and a pre-registered controlled clinical trial.

Mental health improvements:

  • Anxiety (GAD-7): Reductions equivalent to months of traditional therapy

  • Depression (PHQ-8): Effects comparable to one-on-one treatment

  • Wellbeing (ONS-4): 2 WELLBYs gained per participant over 4 years 

  • Proven to be to be as effective as 1:1 psychotherapy (Hofmann et al., 2012)

Real-world impact:

  • Participants gained 6 hours of productive time per week—1.5 hours of additional work time plus 18% productivity increase during working hours

  • In LMICs, this translates to ~2 months of additional income per participant, often covering school fees, healthcare, or essentials

Engagement:

  • <15% dropout rate - as low as the best one-on-one therapy

  • 90%+ want follow-up sessions

  • 80%+ recommend to others

Institutional recognition

  • EA Infrastructure Fund 2025 Featured under "Recent grants with outstanding outcomes"

  • Giving What We Can 2023/24 Featured charity 

  • Clearer Thinking 2022 Funding winner, in the top 4% of 630 applicants

Support our work

Your donation today means support for thousands tomorrow.

Let’s close the global mental health gap together.

Donate now

Frequently asked questions

    • Human oversight: Clinical psychologists design all content and supervise facilitators

    • AI limitations: AI handles information, logistics, and routine support only—not therapy itself

    • Crisis protocols: AI immediately escalates concerning situations (suicidality, abuse) to human experts

    • Continuous monitoring: Every session gets feedback; program leads review patterns weekly

    • Proven track record: Our current human-intensive version maintains 85% retention and strong outcomes

  • We're complementary to in-person programs, not replacing them:

    • We reach populations with barriers to in-person access (rural, mobility-limited, stigma-concerned)

    • We serve countries where StrongMinds and similar programs aren't active (Nigeria, Ghana, etc.)

    • We target younger, educated digital natives who prefer online engagement

    • Both approaches are needed to close the treatment gap

  • Critical difference: Human connection

    Chatbots fail because people use them alone. Our program succeeds because:

    • Small groups create accountability and belonging

    • Trained facilitators guide sessions and build relationships

    • Peer support continues between sessions

    • AI enhances the experience but doesn't replace human connection

    Think of AI as the administrative assistant, not the therapist.

  • We would need to pursue alternative funding sources

  • With successful Phase 1 results:

    • Apply for Phase 2 funding 

    • Begin LMIC adaptation and early adopter pilot

    • Target launch of scaled program by end of Year 1

    • Publish findings to help other organizations adopt similar approaches

    If results don't meet targets:

    • Iterate on design based on learnings

    • Share transparent findings with mental health community

    • Pivot to most promising components or return remaining funds