//Case studies
Woa (WooPro).
A mobile app backend for therapists — 27 workflows and 21 REST endpoints in production. Migration of a therapeutic AI assistant from a Telegram bot to a professional mobile application.
- Industry
- Mental-health app
- Timeline
- 2025
- Outcome
- 27 workflows · 21 REST endpoints in production
The problem
Migrating a therapeutic AI assistant from a Telegram bot to a professional mobile application (Flutter) with a modern REST backend.
What I built
A REST backend on n8n + Supabase:
- 21 endpoints, JWT middleware as a reusable subworkflow,
- RAG per therapy method,
- a two-agent system (intent classifier + therapeutic agent),
- audio transcription (Whisper), the PostgREST pattern, RLS,
- a full E2E test suite and automated API documentation.
The outcome
A production deployment with no critical bugs, delivered on time. The reusable architectural patterns developed here — JWT middleware, data access, file processors — are an example of production engineering discipline in the agentic model.