AntarLens
AI Cognitive Health Platform
A deployed full-stack AI platform that tracks focus, mood, energy and sleep, then predicts a user's cognitive patterns from their own data via a 30-second daily check-in.
- Next.js
- React
- Supabase (Postgres, RLS)
- OpenAI API
- Vercel
- Check-in
- 30 sec
- Signals
- 4
- Isolation
- RLS
- Status
- Live
- 01
Architected the data model and auth on Supabase row-level security, isolating user records at the database layer.
- 02
Built a privacy-first design with encryption at rest and full user memory controls.
- 03
Predicts cognitive patterns from a user's own history rather than population averages.
Security at the database layer
Auth and the data model were designed together around Supabase row-level security. User isolation is enforced by Postgres policies rather than by application code — so a bug in a route handler can't leak another user's rows, because the query itself is scoped before it ever returns.
Privacy as a product decision
Cognitive health data is about as sensitive as personal data gets. The platform encrypts at rest and gives users full control over their stored memory — including the ability to inspect and delete what the system retains about them.
Low-friction input, personal baseline
The entire input surface is a 30-second daily check-in, because adherence is the real constraint on any self-tracking product. Predictions are drawn from each user's own accumulated history, so the baseline is personal rather than a population average that may not describe them at all.