NutriWise

An AI-powered nutrition web app that scans restaurant menus, understands your health needs, and suggests tailored meals with regional options complete with visuals and smart meal planning.

Comment

🥗 NutriWise – Your Personalized AI Health Companion

🚀 About the Project

NutriWise is a modern AI-powered web application designed to revolutionize personal nutrition and health tracking. It acts as your personal assistant, helping you build meal plans, track your macros, analyze restaurant menus, and generate grocery lists—all based on your health goals, dietary restrictions, and regional cuisine preferences.

🌱 What Inspired Us

We noticed a gap in the current health tech market—while there are many apps offering calorie tracking or generic diet suggestions, few consider individual goals, regional food diversity, or AI-driven personalization. With India being home to a variety of rich cuisines, we felt the need for a tool that could personalize nutrition for everyone, from Maharashtra to Manipur. That’s what sparked the creation of NutriWise!

🛠️ How We Built It

We used:

  • Supabase for authentication, user data storage, and dynamic content management.
  • Google Gemini 1.5 Flash for intelligent conversation and meal planning.
  • Google Imagen 2 for AI-generated food imagery from textual menus.
  • Vite + React + TailwindCSS to build a fast and sleek frontend.
  • Hyprland + EndeavourOS as our development environment for optimal performance.

We also integrated real-time menu photo analysis via Gemini Vision (deprecated), and later Imagen2 for rendering food dish previews based on AI meal suggestions.

🧠 What We Learned

  • Implementing LLM-generated responses into a structured web UI is more nuanced than expected—prompt engineering plays a massive role in the quality of output.
  • Fine-tuning user experience around AI latency, feedback loops, and nutrition science was key to making NutriWise feel truly "smart."
  • Integrating image generation required understanding Google Cloud's Vertex AI, secure service credential management, and fallback handling.

🧗 Challenges We Faced

  • API Deprecation: Gemini Pro Vision was deprecated during development. We had to quickly switch to Imagen 2 and re-architect image generation workflows.
  • Stack Depth Issues on Supabase: Large updates caused stack overflow errors due to Postgres limitations. We optimized schema designs and request structures.
  • Regional Cuisine Diversity: Creating inclusive food mapping across all Indian states while maintaining macro balance was a tough nut to crack.
  • Deployment Configuration: Netlify’s environment handling required careful management of sensitive service accounts and image-serving constraints.

💡 What Makes NutriWise Unique

  • Personalized AI meal planning tuned to your macro goals.
  • Menu Analyzer that reads food images or text and breaks them into trackable nutrition.
  • State-wise Indian cuisine support for hyper-localized recommendations.
  • Image generation for meals to visualize suggestions before cooking or ordering.
  • A sleek, gamified progress tracker with user-centric design.

NutriWise isn’t just another meal planner—it’s a nutrition intelligence platform built for modern lifestyles. We’re excited to keep evolving it for global users with a local touch.

Built With

  • bolt
  • framer-motion
  • google-gemini-1.5-flash
  • google-imagen-2-(vertex-ai)
  • lucide-icons
  • netlify
  • postgresql
  • react
  • supabase
  • tailwind-css
  • vite
Share this project:

Updates