State-Aware Adaptive Fitness

Burnout builds silently — poor sleep, skipped workouts, low energy. Existing fitness apps assume you are motivated, and no tool connects mental state to physical capacity. Wellness apps address mental health in isolation from physical habits, while workout apps deliver the same plan regardless of sleep or stress.

A web app with two integrated modules sharing one LLM backbone — a Burnout Tracker and an Adaptive Workout Planner. A weekly 2-minute check-in (sleep, stress, energy, social, enjoyment) feeds an LLM that analyzes multi-week trajectories to output a Burnout Risk Score and micro-interventions. The score auto-adjusts workout difficulty in real time.

• Complete a weekly 2-minute check-in (sleep, stress, energy, social, enjoyment)
• Receive a Burnout Risk Score with micro-interventions
• Configure a workout (muscle group, time, difficulty, equipment)
• LLM generates a structured plan with GIF demos, auto-scaled if burnout is elevated

• Burnout analysis — LLM analyzes multi-week check-in trajectories
• Adaptive workout generation — produces structured JSON plans tied to burnout state
• Real-time difficulty adjustment — the burnout score modifies the generated plan

• Frontend: React + Tailwind
• Backend: Python FastAPI
• LLM: Claude API (structured JSON outputs)
• Exercise data: ExerciseDB API
• Hosting: Vercel
• Vibe coding: Claude Code, Cursor, VS Code Copilot