Imperfect Coach

Inspiration

"What if a routine could adapt to imperfect days instead of punishing them?"

It started on New Year’s Day. My cousin and I sat down with fresh motivation and a shared goal: this year, we wanted better routines. Better sleep, more structure, healthier meals, and time for things that actually mattered to us. We planned our days, set goals, and genuinely believed we’d stay consistent.

A few days later, reality hit. Late nights, skipped workouts, missed meals, and that familiar feeling of guilt. The hardest part wasn’t falling off the routine, it was how quickly it started to feel like failure. Every tool we tried assumed consistency. Miss one day, and it felt like starting over. That’s when we realized something important: the problem wasn’t discipline; it was the absence of compassion.

What it does

Imperfect Coach is a comprehensive AI-powered wellness companion that focuses on the psychological "pivot" rather than the "red X" of failure.

Empathetic AI Conversations: A chatbot interface that remembers context, providing warm support that validates emotions before offering gentle suggestions.
Voice-Enabled Support: Integrated with [suspicious link removed], allowing users to hear their coach’s guidance, creating a personal and accessible human connection.
Adaptive Daily Planning: Generates a balanced plan integrating meals, water, exercise, and "Passion Blocks" based on user insights and body metrics.
Focus Area Tracking:

Sleep: Logging bedtime and wake times to visualize patterns.
Fuel: Personalized meal suggestions with caloric estimates based on age and weight.
Passion: Tracking time spent on creative joy like dancing or coding.

How we built it

We utilized a high-performance stack to ensure the "vibe" of the app stayed fluid and responsive.

Frontend: Built with React 18, Vite, and TypeScript for a responsive architecture.
Styling: Implemented a modern "glass-morphism" design system using Tailwind CSS.
Backend: Created a RESTful API using FastAPI (Python) to handle reasoning and audio generation.
AI Intelligence: Integrated llama model via groq API for reasoning and the ElevenLabs Turbo v2 model for low-latency voice synthesis.

Challenges we ran into

Contextual Depth: Maintaining a "friend-like" memory during a session required careful design of the conversation history to ensure the AI stayed relevant without becoming overwhelmed.
CORS & Connectivity: Setting up secure communication between the Vite frontend and FastAPI backend required precise middleware configuration to handle cross-origin requests during rapid development.
State Persistence: Managing user data across authentication, profile setup, routine, and preferences while ensuring data persists correctly in localStorage required careful state management.

Accomplishments that we're proud of

-Seamless AI Intelligence: Successfully integrated llama model with groq api with a system that automatically detects and routes to the most capable models, ensuring a robust user experience.

Holistic Personalization: Created a system where meal suggestions adapt to body metrics and daily plans integrate seamlessly with a user’s unique routine and "Passions."

-Voice-Enabled Connection: Successfully implemented text-to-speech functionality that makes coaching more personal and accessible.

-Philosophy-Driven UX: We translated the "progress over perfection" philosophy into a UI that celebrates small wins rather than highlighting failures.

-Context-Aware Interactions: Built a chatbot that remembers conversation history within sessions, creating more natural and meaningful dialogues.

What we learned

-The Value of Resilience in APIs: We learned the importance of implementing robust error handling and fallback logic when working with third-party AI services.

-User-Centered AI Design: We discovered how to use personal data (body metrics and routines) to move beyond generic responses and create truly personalized coaching.

-Modern Full-Stack Mastery: Gained deep experience in connecting React/Vite frontends with FastAPI backends, mastering state management, and handling environment security.

-Effective State Patterns: Discovered how to manage complex application states across authentication, profiles, and daily habit tracking efficiently.