Clarity: AI Coaching That Calls You Back

I built Clarity as a solo project for the RevenueCat Shipyard hackathon (Simon’s brief).

The problem I wanted to solve: high-quality coaching is powerful, but expensive and hard to access consistently. Most AI tools also feel too technical and overwhelming for everyday users.

Clarity is a calm, mobile-first coaching app where you can pick a coach, start a session instantly, and get clear next steps. The key difference is that Clarity is not only reactive chat. It can also initiate voice coaching calls and follow-up calls, so accountability happens in real life, not just inside a chat window.

How I built it

I built Clarity with Expo/React Native, Supabase, and RevenueCat, with a full coaching loop:

  • Coach selection by thinking mode (career, mindset, finance, creativity, relationships, productivity)
  • Real-time text coaching and voice sessions
  • AI-generated post-session summaries and action items
  • Memory/context so coaching improves over time
  • Subscription flow powered by RevenueCat (free-to-premium path)

Safety and reliability

I used a fine-tuned coaching model setup and added guardrails for safety and quality:

  • Behavioral guardrails for coaching tone, boundaries, and structured responses
  • Voice call and chat session guardrails
  • Auth + row-level security on user data
  • Controlled follow-up scheduling to avoid spammy behavior

Challenges

The hardest part was making proactive voice calls reliable end-to-end on mobile, especially iOS VoIP + CallKit + realtime audio. I also had to balance a premium experience with a UX that stays simple, calm, and approachable.

What I learned

I learned that great AI coaching is not just model quality. It’s product design, safety guardrails, and timing. The best experience is one that feels human, clear, and useful from the first tap.

Built With

Share this project:

Updates