← Projects

Product

Equi

AI-driven, research-based bipolar disorder monitoring and stabilisation app for iOS and Android.

Mar 2026 – present · Built end to end · an EquiLabs product

  • 20screens shipped
  • 15+clinical features
  • Betalaunch stage

Problem

Over 6 million Americans live with bipolar disorder, yet most go years without a correct diagnosis, consistent monitoring, or care that understands the full picture of their cycles. General wellness apps track mood on a generic 1–5 scale and miss the clinical reality of cycling between mania and depression.

What I built

A clinical-grade monitoring system that fits in a pocket:

  • Cycle tracker: separate mania, depressive and mixed tracking with symptom checklists, shown as a 90-day wave graph with AI-detected patterns.
  • AI wellness reports: weekly and monthly summaries correlating sleep, mood, journal tone, activities and heart rate, generated by Llama 3.1 70B through Groq’s zero-retention API and exported as PDFs for psychiatrists.
  • Notion-style journal: a block editor where each entry is auto-tagged with cycle phase, sleep and activities.
  • Bipolar workbook: an evidence-based CBT/DBT workbook with 7 sections and 49 prompts.
  • Crisis mode: one tap from anywhere to reach emergency contacts, crisis lines and grounding tools.
  • Psychiatrist marketplace: booking and prescribed activities, through an optional clinical layer.
  • AI-moderated community: an anonymous community with no likes, moderated by Perspective API and Llama 3.2 3B.
  • Wearable sync (Apple HealthKit, Google Fit), medication adherence, social-rhythm tracking and a relapse signature builder.

Architecture

React Native (Expo SDK 55) with Expo Router and NativeWind on the client; Supabase (Postgres, Auth, Realtime) as the backend; Zustand for state; expo-sqlite with append-only versioned migrations for offline use; Sentry for crash reporting; Stripe for donations; Calendly for booking. The AI layer has a self-hosted fallback (Ollama + BioMistral) planned for full HIPAA compliance.

Part of EquiLabs

Equi is the first product from EquiLabs, my AI-native technical partner studio. See the EquiLabs venture page.

Principles

Privacy first: data is never sold, and AI runs on zero-retention APIs. The app is donation-funded, so donors sponsor access for people who need it.

Stack