FDA Breakthrough Device Designation for adolescent depression (2022). Deployed across NHS ICS regions in UK, EAP partnerships in US enterprises. The reference design for “clinical SaMD path done correctly.”
The FDA wellness-vs-medical line decides what you can ship. Solo founders can win on the wellness side.
AI mental health chat is the most ethically-loaded track in the 2026 health atlas. Wysa runs $100M+ ARR estimated across employer and NHS contracts with FDA breakthrough designation for adolescent depression; Earkick’s Panda is the iOS top-grossing AI wellness app at $11M+ ARR; Replika and Character.ai are post-lawsuit case studies in what happens when you build emotional AI without guardrails — Replika faced an EU enforcement order, Character.ai is in active wrongful-death litigation as of 2025 with at least three suits filed. The DOJ’s Cerebral settlement in 2024 added a new layer: $7M+ to settle illegal Adderall prescribing practices, proving that telehealth + AI scale becomes an enforcement target the moment it crosses into clinical territory. The honest read: this is a solo-viable track only on the wellness side. The moment your product diagnoses depression, recommends medication, or claims clinical effectiveness, you have crossed into FDA SaMD (Software as a Medical Device) territory — and that’s a $5M+ clearance pathway. The wedge that works: AI journaling, mood tracking, CBT exercises, meditation coaches, sleep companions, with clear “not a substitute for professional care” framing. Solo founders ship $5K-$50K MRR routinely on this side.
FDA Breakthrough Device Designation for adolescent depression (2022). Deployed across NHS ICS regions in UK, EAP partnerships in US enterprises. The reference design for “clinical SaMD path done correctly.”
Panda is the top-grossing AI mental wellness app on iOS. Strictly wellness positioning, no medical claims. Proof that solo+small-team can build a $10M+ ARR consumer business without FDA registration.
Andrew Ng-backed, peer-reviewed CBT chatbot. In active FDA clinical trial roadmap. Slower than Wysa but the clinical purist’s reference — what “therapy chatbot done rigorously” looks like.
Pioneer of companion AI. Italy banned to minors 2023, EU regulator enforcement, pivoted away from sexual roleplay. Still profitable, still warning lesson #1 for any “always agreeable” AI design.
Wrongful-death litigation from teenage user suicide cases. Google paid licensing fee + hired the team, leaving the company shell to handle litigation. The reference case for “build emotional AI without guardrails = existential risk.”
B2B mental health benefits with AI triage layered on top of human therapist network. Proof that the durable money is at the employer-payer-clinician triangle, not in solo consumer apps.
Meditation incumbents added AI chat features in 2024-25 (Headspace AI, Calm AI). Existing distribution + B2B EAP contracts mean new AI-only entrants face a steep gradient.
DOJ settled with Cerebral for illegal Adderall and Xanax telehealth prescribing practices in 2024. The cautionary tale every founder of “AI + telehealth scripts” must know cold.
Journaling, mood tracking, CBT exercises, meditation coaches, sleep companions — none require FDA clearance. Wellness positioning + clear “not a substitute for professional care” framing is solo-viable up to $1M+ ARR.
If you want to build clinical SaMD, an LMFT, LCSW, PsyD, or MD co-founder isn’t optional — it’s the FDA pathway, IRB submission, and credibility moat. Solo engineers don’t pass FDA breakthrough designation review.
Suicidal ideation, self-harm, abuse disclosures — these must route to 988 (in US) or local crisis resources before the AI says anything else. This is non-negotiable in 2026 and is what the Character.ai lawsuits will codify into standard care.
That is the textbook definition of practicing therapy without a license, plus exposure to wrongful-death liability (see Character.ai). Every state bar of psychology + APA position is firm: AI cannot be primary therapy. You will lose this case.
FDA SaMD, HIPAA, state telehealth practitioner laws, FTC AI chatbot rulemaking, COPPA (if anyone under 13), wrongful-death duty of care — minimum five overlapping regulators. Pre-launch $25K legal review is the cheapest insurance you’ll buy.
Replika+Character.ai showed that “maximally engaging companion AI” produces measurable harm at scale. EU and state AGs are coming for this design pattern. Maximizing engagement on emotionally vulnerable users will eat your company.
Founder + licensed clinician (LMFT/LCSW/PsyD/MD), patient-data fluent
Solo founder with product taste; willing to live the wellness boundary
Founder with employer-benefits or clinician-network relationships
Licensed therapists, psychiatric MDs, and digital-health PMs who’ve filed FDA 510(k) or De Novo submissions own the clinical lane. Wysa’s founder Jo Aggarwal is a former Fortune 500 leader; the pattern repeats — domain + product taste wins clinical SaMD.
Mid-career clinicians, healthtech operators, and behavioral-health PMs carry the trust signals + employer relationships needed to land the durable B2B contracts. Spring Health and Lyra were founded by exactly this profile.
If you already run a mental-health-adjacent newsletter, TikTok, or YouTube, the wellness-app path is yours. Stay disciplined about wellness positioning. Earkick proves a solo brand can do $11M ARR without ever crossing the clinical line.
5 min · 12 questions · Free · Get your archetype + top 3 matching tracks
Take the quiz →