🧠 MoodBite.Ai (v3.0)

The Offline-First Sensory Food Interface

MoodBite is a sophisticated emotional regulation system that uses culinary intervention to balance human mood states. Unlike generic AI wrappers, it runs on a custom Deterministic Emotional Vector Machine (EVM), allowing for high-precision, offline emotion detection and reasoning.

🌐 Live Demo: https://moodbiteai.vercel.app/

---

🏛️ Architectural Diagram

The core of MoodBite v3.0 relies on an "Expert Panel" architecture where multiple specialized logic modules vote on the user's state before a final "Judge" determines the emotional vector.

graph TD
    User([User]) -->|Text/Voice| Input
    User -->|Fridge Photo| Vision[Gemini Vision API]
    
    subgraph "Sensory Layer"
        Input --> Context[Context Expert]
        Input --> Lexicon[Lexicon Expert]
        Input --> Conflict[Conflict Expert]
    end

    Vision -->|Ingredients| FoodEngine

    subgraph "Decision Layer (The Judge)"
        Context & Lexicon & Conflict --> EVM{EVM Vector Engine}
        EVM -->|Valence, Arousal, Control| State[Emotional State]
    end

    subgraph "Action Layer"
        State --> Strategy[Strategy Selector]
        Strategy --> FoodEngine[Food Recommendation Engine]
        FoodEngine -->|Database + Fridge Match| Response
    end
    
    Response --> UI[Final UI]

Loading

---

🎨 Design Model

MoodBite follows a 3-Layer Sensory Design Philosophy:

1. The Sensory Surface (UI):

   • Glassmorphism: Built with backdrop-filter, bg-white/6, and deep gradients to mimic a futuristic dashboard.
   • Dynamic Backgrounds: The environment shifts colors (Red for Anger, Blue for Grief) based on real-time vector analysis.
   • Interactive Input: Supports Voice Command (Web Speech API) and Vision (Upload Fridge Photos).

2. The Expert Core (Logic):

   • Offline-First: 90% of the logic (Emotion detection, Reasoning, Food matching) runs locally without external API calls.
   • Deterministic: Same input = Same output. No hallucinations.
   • Explainable AI: The system generates a "Chef's Note" explaining why a specific food regulates a specific mood.

3. The Vision Layer (Integration):

   • Integrates Google Gemini 1.5 Flash to analyze refrigerator contents and cross-reference them with the emotional strategy.

---

🛠️ Implementation & Tech Stack

• Framework: Next.js 15 (React 19)
• Styling: Tailwind CSS v4 + Framer Motion
• AI (Vision): Google Gemini 1.5 Flash
• AI (Core): Custom JS-based Expert System (No LLM dependency for logic)
• Image Gen: Pollinations.ai (Real-time food visualization)

Directory Structure

├── components/       # UI Building Blocks (ModernInputBar, HeroFoodCard)
├── evm/             # The Brain (Expert Vector Machine)
│   ├── dictionaries.js  # 28-Emotion Lexicon
│   ├── experts.js       # Signal Processors
│   ├── judge.js         # Vector Math & Decision Logic
│   └── foodEngine.js    # Strategy & Recipe Database
├── pages/           # Routes (Home, About, API)
└── public/          # Static Assets

---

🔄 Deployment Workflow

The project is configured for seamless deployment on Vercel.

GitHub Actions Workflow (CI/CD)

While Vercel handles deployments automatically, here is the logic flow for a standard CI pipeline:

name: MoodBite Build Pipeline
on: [push]

jobs:
  build:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - name: Install Dependencies
        run: npm ci
      - name: Lint Analysis
        run: npm run lint
      - name: Production Build
        run: npm run build
      - name: Test Offline Engine
        run: node evm/test_suite.js

---

🚀 Getting Started

1. Clone the Repo

   git clone https://github.com/singhcod3r/MoodBite-Version-2.0.git
   cd MoodBite-Version-2.0

2. Install Dependencies

   npm install
   # or
   yarn install

3. Set Environment Variables Create a .env.local file:

   GEMINI_API_KEY=your_gemini_key_here

4. Run Development Server

   npm run dev

5. Open Browser Visit http://localhost:3000.

---

© 2025 MoodBite Project

Designed & Engineered by Ayush Singh.