An e-commerce furniture store with an intelligent AI chatbot assistant built with React, Node.js, LangChain, LangGraph, and Google's Gemini AI. The website features a modern dark-themed UI with neon green accents and uses MongoDB Atlas for vector search capabilities and conversation persistence.
- Intelligent Furniture Search: Vector-based search through furniture inventory using Google's text-embedding-004 model
- Conversational AI: Powered by Gemini-1.5-Flash for natural language understanding
- Persistent Conversations: Chat history stored in MongoDB with conversation threading
- Fallback Search: Automatic text-based search when vector search fails
- Rate Limiting: Built-in exponential backoff for API rate limiting
- RESTful API: Express.js server with CORS support
- Modern E-commerce UI: Dark theme with neon green colors
- Interactive Chat Widget: Floating chat bot with smooth animations
- Real-time Chat: Seamless conversation flow with the AI assistant
- React 19 with TypeScript
- Vite
- React Icons
- Axios
- CSS3
- Node.js + TypeScript + Express.js
- LangChain + LangGraph for AI workflow management
- Google Gemini AI for language processing
- MongoDB Atlas with vector search capabilities
- User Input: Receive customer message via React frontend
- Tool Decision: Determine if inventory search is needed
- Vector Search: Query furniture database using embeddings
- Text Fallback: Use regex search if vector search fails
- Response Generation: Create helpful, contextual response
- State Persistence: Save conversation to MongoDB
- Frontend Update: Display response in chat widget
- Rate Limiting: Exponential backoff for API limits (429 errors)
- Authentication: Graceful handling of API key issues (401 errors)
- Database Errors: Fallback to text search when vector search fails
- Empty Inventory: Informative messages when no items found