🧠 AI-Native Nutrition Co-Pilot

Make sense of food ingredients at the moment decisions matter.

🔗 Live Demo website : https://encode-hackathon-nine.vercel.app

---

🚀 Problem We’re Solving

Food labels are built for regulatory compliance, not human understanding. Consumers struggle with:

• Long, confusing ingredient lists
• Complex chemical names
• Conflicting health advice

Existing tools often just dump ingredient info instead of explaining meaning. This project answers the hackathon challenge to design an AI-Native experience where AI isn’t just a feature—it is the interface.

---

🎯 What is AI-Native Here?

According to the hackathon brief, AI-native systems must infer intent, reduce cognitive effort, and act as a co-pilot. Our system delivers:

• ✅ SAFE / CAUTION / AVOID verdicts
• ✅ Why it matters: Contextual explanations
• ✅ Tradeoffs: Balanced health views
• ✅ Uncertainty explanation: Honest AI reasoning
• ✅ Highlighted risky ingredients: Visual focus on what matters

---

🛠 Architecture & Tech Stack

Frontend

• Framework: React + Vite
• Styling: Tailwind UI
• Auth: Secure token storage
• UX: AI-native immersive UI theme

Backend

• Environment: Node + Express
• Database: MongoDB (Atlas)
• Auth: JWT Authentication
• LLM: OpenAI GPT-4o-mini for reasoning

Database Schema

MongoDB stores:

• User credentials
• User persona (gym-goer, diabetic, allergy-prone, etc.)
• Persistent personalization

---

👤 How It Works (User Flow)

1. Identity: User signs up / logs in via secure JWT flow.
2. Input: User enters ingredients (image/text input ready in future roadmap).
3. Personalization: User defines a persona (e.g., "diabetic" or "vegan").
4. Reasoning: AI evaluates the list based on the specific persona.
5. Verdict: Returns an explainable AI judgement with a theme color (Green/Yellow/Red).

---

🔐 Authentication Flow

• Signup/Login returns a JWT.
• The api/judge route is protected.
• Middleware verifies the token on every request.
• The Persona persists per-user, allowing for a "Memory + Personal Context" experience where future queries automatically adapt.

---

📊 Sample Output

The system provides structured JSON responses to drive the immersive UI:

{
  "verdict": "CAUTION",
  "short_reason": "Contains high sugar + whey risk for lactose sensitive users",
  "detailed_reason": "...",
  "ui_theme": "yellow",
  "highlighted_ingredients": ["sugar", "corn syrup", "whey protein"],
  "uncertainty_note": "Limited data about user health"
}

📦 Installation

Backend

Navigate to the backend directory and install dependencies:

cd backend npm install touch .env

Add the following to your .env file:

MONGO_URI=your_mongodb_uri JWT_SECRET=your_secret_key OPENAI_KEY=your_openai_api_key

npm run dev

Frontend

cd frontend npm install npm run dev