🌍 VoyageAI — AI-Powered Travel Planner

VoyageAI is a full-stack AI travel planner that generates complete personalised trip plans using Groq (LLaMA) and Google Gemini AI. Enter your source city, destination, budget, and trip type — get a full day-wise itinerary, hotel picks, food guide, budget breakdown, packing list, and emergency contacts instantly.

Live Demo: voyage-ai-2.onrender.com

---

✨ Features

🗺️ Trip Planning

• 📅 Day-wise Itinerary — Morning, Afternoon, Evening plans for every day
• 🏆 Trip Score Card — AI rates your trip on Adventure, Romance, Food, Value & more
• 💰 Budget Assessment — Tells if budget is sufficient with min & comfortable estimates
• 💱 Auto Currency Detection — Detects destination currency, shows INR equivalent for international trips
• 🚗 Smart Travel Mode — Disables modes not available for your route (e.g. no flight for Delhi→Agra)
• 🏨 Hotel Recommendations — Tailored to trip type (Solo / Couple / Family / Friends / Honeymoon)
• 🍛 Food Guide — Must-try local dishes and restaurant picks
• 📸 Places to Visit — Top attractions with entry fees
• 🎒 Interactive Packing List — Tap to check off items
• 🆘 Emergency Contacts — Local police, ambulance, embassy
• 🖨️ Print / Save as PDF — Full plan printable
• 🕒 Recent Trips — Last 5 plans saved locally

🤖 AI-Powered Smart Features

• 🔍 Semantic Destination Search — Natural language search using TF-IDF vectorization with cosine similarity over a curated travel knowledge base. Understands mood, weather, and season queries like "rainy misty peaceful", "cold snowy romantic", "sunny beach party"

• 📊 Budget Optimizer — Smart budget allocation by category (travel / hotels / food / transport / activities) with health score (0–100), per-person breakdown, and saving tips — personalised by trip type

• 🗺️ Route Optimizer — Multi-city route planner using Nearest Neighbor + 2-opt local search to minimize total travel distance. Detects international routes and restricts impossible modes (no bike/train for Delhi→London)

• ✨ Personalized Recommendations — Similar destination suggestions scored on trip-type affinity, budget match, and semantic similarity

• 🧠 Grounded AI Generation — Trip plans are grounded using semantic retrieval over a curated travel knowledge base (TF-IDF vectorization + cosine similarity). Relevant destination facts — best season, budget level, key highlights — are injected into the AI prompt to reduce hallucination and improve accuracy

• 💬 AI Travel Assistant Chatbot — Context-aware floating chatbot powered by Groq (LLaMA 3.3 70B). Answers real-time questions: "Where should I eat?", "Find ATM", "It started raining", "Shift today's plan"

---

🛠️ Tech Stack

Layer	Technology
Frontend	React 19 + Vite 8
Backend	FastAPI (Python)
Primary AI	Groq API — LLaMA 3.3 70B (fast, generous free tier)
Fallback AI	Google Gemini API (gemini-2.5-flash)
Semantic Search	TF-IDF vectorization + cosine similarity (no external vector DB)
Route Optimization	Nearest Neighbor TSP + 2-opt local search (Haversine distances)
Styling	Pure CSS
Deployment	Render (backend) + GitHub Pages (frontend)

---

🚀 Getting Started

1. Clone the repo

git clone https://github.com/your-username/VoyageAI.git
cd VoyageAI

2. Backend Setup

cd backend
python -m venv venv
venv\Scripts\activate        # Windows
# source venv/bin/activate   # Mac/Linux
pip install -r requirements.txt

Create .env file:

cp .env.example .env

Add your API keys in backend/.env:

GROQ_API_KEY=your_groq_api_key_here
GEMINI_API_KEY=your_gemini_api_key_here

• Groq free key: https://console.groq.com/keys
• Gemini free key: https://aistudio.google.com/apikey

Start the backend:

python -m uvicorn main:app --reload --port 8000

3. Frontend Setup

cd frontend
npm install
npm run dev

4. Open in browser

http://localhost:5173

---

📁 Project Structure

VoyageAI/
├── backend/
│   ├── main.py                   # FastAPI server — all endpoints + AI integration
│   ├── rag_knowledge.py          # Curated travel knowledge base with TF-IDF semantic search
│   ├── budget_optimizer.py       # Smart budget allocation engine
│   ├── route_optimizer.py        # Multi-city route optimizer (Nearest Neighbor + 2-opt)
│   ├── recommendation_engine.py  # Personalized destination recommendations
│   ├── requirements.txt
│   ├── render.yaml               # Render deployment config
│   ├── .env                      # Your API keys (not committed)
│   └── .env.example
├── frontend/
│   ├── src/
│   │   ├── components/
│   │   │   ├── TripPlanner.jsx       # Main form + smart travel mode logic
│   │   │   ├── ResultCard.jsx        # 12-tab results display
│   │   │   ├── SemanticSearch.jsx    # Natural language destination search
│   │   │   ├── BudgetOptimizer.jsx   # Budget allocation visualizer
│   │   │   ├── RouteOptimizer.jsx    # Multi-city route planner UI
│   │   │   ├── RecommendationEngine.jsx  # Similar destinations
│   │   │   ├── TravelChatbot.jsx     # Floating AI travel assistant
│   │   │   ├── Navbar.jsx
│   │   │   ├── Hero.jsx
│   │   │   ├── Features.jsx
│   │   │   ├── Footer.jsx
│   │   │   └── TripPlanner.css       # All component styles
│   │   ├── services/
│   │   │   └── api.js               # All backend API calls
│   │   └── App.jsx
│   └── vite.config.js
└── README.md

---

🔌 API Endpoints

Method	Endpoint	Description
GET	/	Health check
POST	/generate-trip	Generate full AI trip plan (Groq + Gemini fallback)
POST	/search-destinations	Semantic destination search (TF-IDF + cosine similarity)
POST	/optimize-budget	Smart budget allocation by trip type
POST	/optimize-route	Multi-city route optimization (Nearest Neighbor + 2-opt)
POST	/recommendations	Personalized destination suggestions
POST	/chat	AI travel assistant (Groq LLaMA 3.3 70B)

---

🌐 Environment Variables

Backend (backend/.env):

GROQ_API_KEY=your_groq_api_key       # Primary AI — LLaMA 3.3 70B
GEMINI_API_KEY=your_gemini_api_key   # Fallback AI

Frontend (for production deployment):

VITE_API_URL=https://your-render-backend-url.onrender.com

---

Built with ❤️ using React, FastAPI, Groq (LLaMA) and Google Gemini AI