Inspiration

Therapy works — but only one hour a week. The other 167 hours, people are on their own. Sorca was born from a simple idea: what if support didn't end when the session did?

What it does

Sorca is an AI therapy companion built for the gap between sessions. Users can:

  • Have voice sessions with an empathetic AI trained on evidence-based modalities (CBT, ACT, mindfulness)
  • Track mood, sleep and triggers with daily check-ins
  • Complete CBT homework assigned by their therapist
  • Generate PHQ-9 and GAD-7 outcome reports for clinical review
  • Share progress with their human therapist via a supervised dashboard

How we built it

  • Frontend: Next.js 15, React Native (Expo), TypeScript, Tailwind / NativeWind
  • Backend: Node.js, FastAPI, Supabase, PostgreSQL with pgvector for memory
  • AI: Claude API for clinical reasoning, ElevenLabs for low-latency voice, custom RAG over evidence-based therapy protocols
  • Compliance: GDPR-compliant data layer, encryption at rest and in transit, full audit logs for therapist oversight
  • Payments: Stripe subscriptions + therapist seat billing

Challenges we ran into

Building something safe in a mental-health context meant every prompt, every response, every escalation path had to be reviewed. We built a safety classifier that runs on every turn and routes high-risk conversations to crisis resources and human therapists — not "tips."

Accomplishments that we're proud of

  • Deployed with NHS and private UK practitioners
  • Sub-second voice latency end-to-end
  • Real clinical outcomes measured via PHQ-9 / GAD-7 reductions

What we learned

AI in healthcare isn't about replacing clinicians — it's about extending their reach. The therapist stays in charge; the AI just keeps showing up between sessions.

What's next for Sorca

NHS pilots expanding, group therapy mode, and clinician-side analytics so therapists can see exactly what their clients worked on this week — before the next session starts.

Built With

Share this project:

Updates