FoodMind

Explore FoodMind's vibrant homepage with AI meal planning & IoT insights! #FoodMind #AIcooking
Scan food with TensorFlow.js in FoodMind’s scanner—get instant nutrition data! #FoodTech #HealthyEating
3D recipe visualization in FoodMind, powered by Three.js for a tasty twist! #FoodInnovation #TechCooking
Chat with FoodMind’s AI for live cooking tips via IoT sensors! #AIChef #SmartKitchen
Track nutrition with FoodMind’s Recharts dashboard—stay healthy effortlessly! #NutritionTech #MealPlanning
"FoodMind’s PWA in action, offline-ready with live IoT data sync! #ProgressiveWebApp #SmartCooking

FoodMind

FoodMind is a revolutionary nutrition and meal planning Progressive Web App (PWA) that integrates artificial intelligence, real-time IoT sensor data, and a user-friendly interface to transform how people cook and eat healthily. By combining AI-driven meal planning, food recognition, and IoT data from kitchen devices (e.g., temperature, pH sensors), FoodMind acts as a virtual chef and nutritionist, delivering personalized, context-aware cooking guidance and dietary tracking.

Inspiration

The inspiration for FoodMind came from the limitations of static meal planning tools that ignore real-time kitchen conditions and individual dietary needs. We were motivated to create a dynamic solution that leverages IoT data—such as oven temperature or fermentation pH—and advanced AI to provide tailored cooking recommendations. Our vision was to make healthy cooking accessible, engaging, and precise, like having a personal chef in your pocket.

What it does

FoodMind empowers users to cook smarter and live healthier through:

AI-Powered Meal Planning: Generates personalized weekly meal plans and recipes based on user preferences, dietary restrictions, and live IoT data. IoT Integration: Connects with kitchen devices to ingest real-time sensor data (e.g., temperature, pH) via a POST /sensor-data endpoint, offering actionable cooking tips (e.g., "pH 4.5—bake now"). Food Recognition: Uses TensorFlow.js with MobileNet to identify foods from images, providing instant nutritional analysis. Nutrition Tracking: Visualizes calorie and nutrient data with interactive Recharts dashboards. PWA Features: Delivers a mobile-first, offline-capable experience with live gauges, Web Push notifications for sensor thresholds, and an AI chat panel for cooking advice. Community and Support: Includes a forum for sharing recipes, video tutorials, and static pages (Help, About, Contact).

How we built it

FoodMind was built with a modern, performant tech stack optimized for Bolt.new: Frontend: React (TypeScript), Vite for fast builds, Chakra UI for responsive design, Framer Motion for animations, Three.js for 3D recipe visualizations, TensorFlow.js for client-side food recognition, and tsParticles for dynamic backgrounds. Backend: Node.js with Express.js, using Supabase for real-time database, authentication, and Row Level Security (RLS). IoT and Real-Time Sync: REST API (POST /sensor-data) and WebSocket subscriptions for live sensor updates (e.g., temperature, pH), stored in Supabase at /users/{uid}/devices/{device_id}. AI: OpenAI API for recipe generation and cooking tips, with prompts optimized to ≤620 tokens and responses capped at ≤300 tokens to stay within a 2.5 million token budget. Styling: Soft green (#A8D5BA), warm yellow (#FFD700), off-white (#F5F5F5), and dark gray (#333333) color scheme with Open Sans font, featuring glass morphism cards and floating animations. PWA: Service Workers and IndexedDB for offline caching of 24 hours of sensor data, plus Web Push notifications for alerts. Deployment: Hosted on Netlify (frontend) and Render (backend), with CI/CD via GitHub Actions.

Challenges we ran into

Token Budget Management: Optimizing AI prompts to fit within the 2.5 million token limit required careful design of compact templates and retrieval-augmented prompts using precomputed embeddings. IoT Real-Time Integration: Ensuring low-latency WebSocket updates for sensor data while maintaining PWA responsiveness was complex. Cross-Device Compatibility: Supporting camera access for food recognition and AR cooking features across browsers and devices demanded extensive testing. Bolt.new Setup: Configuring Vite proxy and CORS for Bolt.new’s preview URLs to avoid localhost dependencies posed initial hurdles.

Accomplishments that we're proud of

Seamless IoT-AI Synergy: Integrated real-time sensor data with AI recommendations, enabling context-aware cooking advice (e.g., adjusting dough based on pH). Token Efficiency: Achieved ~2,717 AI interactions within the token budget through summarized data and capped responses. Polished PWA: Delivered a mobile-first, offline-capable app with live gauges, 3D visuals, and a cohesive design aligned with FoodMind’s aesthetic. Robust Security: Implemented JWT authentication and Supabase RLS for user-specific data access.

What we learned

IoT Data Handling: Efficiently processing and visualizing real-time sensor data using WebSockets and compact summaries. AI Optimization: Techniques for managing LLM token usage with retrieval-augmented generation. PWA Development: Building offline-first apps with Service Workers, IndexedDB, and Web Push notifications. Security: Best practices for JWT-based authentication and database security with RLS.

What's next for FoodMind

Community Forum: Expand the /forum route for users to share recipes and tips. Video Tutorials: Add step-by-step cooking videos at /tutorials. Expanded IoT Support: Integrate with smart scales, refrigerators, and other kitchen devices. Enhanced AI: Develop more sophisticated models for hyper-personalized recommendations. Monetization: Introduce premium features like advanced analytics and exclusive recipes via Stripe subscriptions.