Personal Fitness Tracker

A personal fitness dashboard that connects to Fitbit for real-time health data and generates science-based performance reports using a multi-agent AI pipeline.

Dashboard

The home view aggregates the latest metrics from Fitbit, recent workouts, and body composition trends into a single glanceable overview.

Body Composition

SECA bioimpedance scan imports with trend analysis, fat-free mass index (FFMI) calculation, and recomp projections. Tracks weight, body fat percentage, and lean mass over time.

Vitals

Real-time Fitbit data including resting heart rate, HRV (RMSSD), VO2 max estimates, sleep scores, steps, and active zone minutes. Each metric is plotted over configurable time periods with 7-day moving averages.

AI Health Reports

Five specialist AI agents (Cardio, Strength, Sleep, Body Composition, Nutrition) analyze health data in parallel, each grounded in sports science research. A summary agent synthesizes their findings into actionable insights with interactive Plotly charts and science-based scoring rubrics.

How it works

• Fitbit Integration -- automatic sync of heart rate, HRV, sleep stages, VO2 max, steps, and active zone minutes
• Multi-Agent Pipeline -- each agent runs independently with its own science framework (Peter Attia, Matthew Walker, Israetel volume landmarks, ISSN protein guidelines)
• Workout Logging -- manual entry plus bulk import from Strong/Hevy with automatic unit conversion
• Scheduled Generation -- Vercel cron jobs trigger weekly (Monday 6 AM) and monthly (1st of month) reports automatically

Stack

Next.js 16, TypeScript, Supabase (auth + Postgres with RLS), Fitbit Web API, Claude/GPT-4o for agent inference, Plotly.js for interactive charts, Tailwind CSS 4, Framer Motion, deployed on Vercel.