Release Coach AI: voice-first emotional coaching sessions
A React and Supabase platform delivering guided voice sessions for emotional reset, with a members area, trainer program, subscriptions and an admin panel behind it.
The product problem
Release Coach AI guides people through a structured method to release stress and regain clarity. The method is spoken, not read: the user states a challenge, rates its intensity, repeats phrases and checks in again until the intensity drops. Turning that into software means a voice session that listens, validates answers and loops correctly, in any language, while keeping each session private and unrecorded.
The delivered scope
The platform is a React and TypeScript single-page app on Supabase, with PostgreSQL, authentication and edge functions behind it. The voice session plays a guided audio prompt for each step and listens through speech recognition. Around it sit a members dashboard, progress tracking, a trainer application and certification program, Stripe subscriptions, affiliate commissions and an admin panel for curriculum, memberships, voices and translations.
A session engine built as a sequence of steps
Each session runs as an explicit sequence of steps, with modes for asking, speaking and waiting for the user to repeat a phrase. The next step depends on the answer: a non-zero intensity sends the user back for another round. Modelling the method this way keeps the conversation predictable and testable. For a similar voice product, defining the steps and branches first avoids building an open-ended chat where a guided flow is needed.
What shipped
A voice-first coaching platform with guided sessions, a members area, a trainer program, subscription billing and an admin panel for content and translations.
Visit the project ↗