'We lack quantitative measures for resilience and health. Our team will build quantitative science-based measures that predict long-term outcomes with the ultimate goal of keeping people healthy. Dr. Michael Snyder, Director of the Center for Genomics and Personalized Medicine at Stanford University'
Aybars Dorman
Gelecek Sistemler Stratejisti & Dorman Review Yazarı
AI Therapy Apps: Conversational Therapy · Companions & Mindfulness · Enterprise Care
'We lack quantitative measures for resilience and health. Our team will build quantitative science-based measures that predict long-term outcomes with the ultimate goal of keeping people healthy. Dr. Michael Snyder, Director of the Center for Genomics and Personalized Medicine at Stanford University'
❓ What You'll Learn
Why did Woebot shut down after raising $124M and securing FDA Breakthrough status?
How did Spring Health hit a $3.3B valuation without ever selling to consumers?
Which states have already banned AI therapy outright?
Why is ChatGPT quietly the largest mental health provider in America by session volume?
Why will at least 25% of US therapists in private practice use an AI co-pilot weekly?
How does an ADHD AI coach become a $19/mo consumer product with no major AI player positioned for it?
Why is mental wellness app retention only 3% at 30 days, and what does that mean for unit economics?
Why does AI lower the cost of disclosure but risk rerouting users away from human help?
💎 Why It Matters
The global AI in mental health market is projected to reach $9.1 billion by 2033 (23.3% CAGR).
1 billion people live with a mental health condition, but the world only has 13 mental health workers for every 100,000 people.
ChatGPT now handles more weekly mental health conversations than the entire licensed US therapist workforce delivers in a year.
The first line of mental healthcare just stopped being human.
The mathematical scarcity of human therapists means care must scale through code.
🔍 Problem
One billion people globally live with a mental health condition.
The world has only 13 mental health workers for every 100,000 people.
The mathematical supply-and-demand ratio for human-only care is permanently broken.
💡 Solution
AI therapy apps deliver evidence-based protocols (CBT, DBT, ACT, motivational interviewing) at a price and scale humans simply cannot match.
They operate dynamically as therapist co-pilots, between-session companions, intelligent journaling apps, and crisis detection layers.
🏁 Players
Conversational AI for Mental Health
Slingshot AI / Ash • Foundation model trained specifically for therapy. The most-watched a16z-backed entrant.
Wysa• Clinically validated AI companion holding an FDA Breakthrough Designation.
Sonia• AI therapist app designed for everyday mental health support.
AI Companion Apps
Replika • The original AI companion, increasingly used for therapeutic support at scale.
Character.AI • AI character chat platform currently facing active mental health litigation.
Pi by Inflection • A highly empathetic AI assistant studied as a viable therapy alternative.
AI Journaling Apps
Rosebud • Therapist-designed AI journal that secured a $6.75M seed round with 500M words logged.
Mindsera • Cognitive performance journaling tailored for high-achieving professionals.
Stoic• A Stoicism-inspired AI mood and daily reflection application.
Employer & Enterprise Mental Health
Spring Health • $3.3B valuation. Fortune 500 employer-distributed care navigation platform.
Lyra Health • Enterprise mental health benefits backed by a massive provider network.
Therapist Infrastructure
SimplePractice • 200,000+ paid users running solo and group therapy practices.
Headway • Insurance-credentialing infrastructure that integrates AI tooling for therapists.
🔮 Predictions by Aybars D.
A mass-casualty regulatory inflection will reshape the category.
With 0.15% of OpenAI's 800M weekly users showing suicidal indicators, major platforms face thousands of highly vulnerable users daily. The Character.AI lawsuits involve confirmed user deaths, and the FTC is actively targeting chatbot operators. Safety protocols will dictate which companies survive the fallout.
Data, not products, will drive acquisitions.
A behavioral health AI startup will be bought strictly for its longitudinal dataset. UHS acquired Talkspace for $835M to capture transactional therapy data alongside a clinician network. Expect Big Pharma (like Lilly or Pfizer) to acquire these platforms to fuel their depression drug pipelines.
Apple will ship integrated mental health AI in the Health app.
Apple's strict privacy positioning fits mental health perfectly. The iOS 26.4 Health app update added mood logging and clinical questionnaires (PHQ-9, GAD-7). Combined with a redesigned Siri, Apple has the technical pieces to dominate ambient health tracking while avoiding the political backlash Meta would face.
☁️ Opportunities
Build a $19/mo ADHD coach and body-doubling AI.
Adult ADHD diagnoses have doubled, yet no major AI player owns this identity-rich market. Bundle morning intention-setting and real-time body-doubling for the 2 AM brain spiral. Pitch directly to ADHD creators on TikTok • 'Always on. No appointments.'
Own AI journaling for high-risk professions.
Quantify ICU nurse turnover ($40K-$60K per nurse) as a budgeted hospital line item. Pitch directly to nursing communities, 'Process your shift before you carry it home.' Targeting active wellness budgets bypasses the saturated consumer app store.
Launch a between-session companion sold to therapists.
The patient uses the app, but the therapist buys the dashboard. Ship CBT thought records, mood tracking, and homework reminders. Keeping the therapist as the clinical owner provides the ultimate regulatory armor.
Sell a $14/mo voice-first sober coach.
Target between-meeting support inside recovery communities. Build a voice-first lane for 2 AM cravings when calling a sponsor isn't possible. Partner with massive networks like The Phoenix to secure instant distribution.
🏔️ Risks
Liability Precedents • The Character.AI lawsuits are setting harsh legal precedents for synthetic empathy before founders can pivot.
Retention Reality • Mental wellness apps lose up to 60% of users in the first month. Clinical claims rarely survive this brutal churn rate.
Data Weaponization • Most consumer apps avoid HIPAA by claiming to be 'wellness' tools. Leaking a journal entry about suicidal ideation is a catastrophic brand risk.
Stigma Reroute • AI makes it easier to disclose trauma to a chatbot, but this might reduce the likelihood of a patient seeking necessary human medical care.
🔑 Key Lessons
Position as augmentation, not replacement. Spring Health hit a $3.3B valuation by routing care, and Wysa pivoted to a gateway model. Augmentation keeps the human in the loop, providing durable regulatory armor.
The generalist LLM is your real competitor. ChatGPT is free, always on, and already trusted. If your app feature fails to beat the basic ChatGPT baseline, cut it from your roadmap immediately.
🔥 Hot Takes
ChatGPT has already won the initial volume war. It handles more sessions than the entire US workforce and will remain the world's largest mental health provider throughout the FDA regulatory cycle. No specialized app can match a trusted, $0 substitute.
The first AI therapy unicorn won't sell to consumers. Built on a B2B distribution model, the buyer is a corporate benefits team, the user calls it 'support', and the brand operates exactly like enterprise SaaS.
😠 Haters
'Users are dying inside chatbots; you're building straight into legal exposure.'
The legal wave is real (FTC orders, Section 230 rulings). However, founders who ship crisis detection as an absolute priority will survive the initial regulatory shock, while negligent platforms will be regulated out of existence.
'Mental health data is uniquely sensitive, and consumer apps escape HIPAA.'
This is precisely why data hygiene is a defensible moat. BetterHelp paid $7.8M for sharing intake data. Building HIPAA-grade infrastructure, even when wellness framing lets you skip it, transforms privacy into a competitive advantage.
'ChatGPT eats this category and doesn't even know it's competing.'
A $0 substitute easily wins the brand war.
The only survival tactic is structural differentiation, domain-specific protocols (DBT, MI), longitudinal clinical memory, and hard outcomes that users can hand directly to their insurer.
'Most users drop off in weeks; the clinical claims are useless against churn.'
D2C standalone apps face a 97% churn rate. The fix is abandoning D2C entirely. Shift to between-session augmentation where the human therapist ensures continuity, or pivot to AI scribe tools for practitioners.
'AI therapy reroutes users away from real human care.'
This is a valid clinical concern. AI relationships can replace human ones. Defensible apps must optimize for the 'warm handoff' to human clinicians during a crisis, rather than running a social media playbook to maximize screen time.
(https://doi.org/10.1056/AIoa2400802) • Peer-reviewed randomized trial of a generative therapy chatbot, reframing what 'evidence' means for this class.
Ask me anything about Aybars Dorman's work, projects, or advisory services. I can help with questions about green energy, AI, digital economies, and longevity.
Bu web sitesi, kullanıcı deneyimini iyileştirmek ve anonim analitik verileri işlemek için gerekli çerezleri kullanır.
