PartSelect Chat Assistant - Instalily AI Case Study

Developer: Bob Tianqi Wei
Date: June 2025

A React-based chat assistant for PartSelect e-commerce website, specializing in refrigerator and dishwasher parts. Features DeepSeek AI integration, product search, compatibility checks, installation guides, and troubleshooting support.

🎥 Demo Video

https://youtu.be/WIz0ok_3UnQ

🚀 Core Features

• AI-Powered Chat: DeepSeek language model integration for intelligent responses
• Product Search: Search for refrigerator and dishwasher parts by number or description
• Compatibility Checks: Verify part compatibility with appliance models
• Installation Guides: Step-by-step installation instructions with visual guides
• Troubleshooting Support: Common appliance issue solutions
• Modern UI/UX: Responsive React interface with real-time chat
• Vector Database: Semantic search using Pinecone (optional)

🛠️ Technology Stack

• Frontend: React.js, CSS3
• Backend: Node.js, Express.js
• AI: DeepSeek API
• Vector Database: Pinecone (optional)
• Images: SVG graphics for products and installation guides

📁 Project Structure

System architecture flowchart of the PartSelect Chat Agent

ChatAgent_PartSelect/
├── src/
│   ├── components/
│   │   ├── ChatWindow.js          # Main chat interface
│   │   ├── ProductCard.js         # Product display component
│   │   └── CompatibilityChecker.js # Compatibility verification
│   ├── data/
│   │   └── sampleProducts.js      # Sample product database
│   ├── services/
│   │   └── vectorDB.js            # Vector database service
│   └── api/
│       └── api.js                 # API client
├── public/
│   └── images/products/           # Product images and installation guides
├── server.js                      # Express backend server
├── config.js                      # API configuration
└── package.json

🔧 Installation & Setup

Prerequisites

• Node.js (v14 or higher)
• npm package manager
• DeepSeek API key
• Pinecone API key (optional)

Quick Start

1. Clone and install

   git clone https://github.com/bobtianqiwei/ChatAgent_PartSelect.git
   cd ChatAgent_PartSelect
   npm install

2. Configure API keys

   Copy config.example.js to config.js:

   cp config.example.js config.js

   Update config.js with your API keys:

   module.exports = {
     DEEPSEEK_API_KEY: 'your-deepseek-api-key-here',
     DEEPSEEK_API_URL: 'https://api.deepseek.com/v1/chat/completions',
     PINECONE_API_KEY: 'your-pinecone-api-key-here', // Optional
     PORT: 3001,
     NODE_ENV: 'development'
   };

3. Start the application

   Start the backend server:

   node server.js

   Start the frontend (in a new terminal):

   npm start

   The application will be available at:

   • Frontend: http://localhost:3000
   • Backend API: http://localhost:3001

📚 API Endpoints

Core Endpoints

• POST /api/chat - Send chat messages to AI
• GET /api/products - Search products
• GET /api/compatibility/:partNumber - Check part compatibility
• GET /api/installation/:partNumber - Get installation guide
• GET /api/troubleshooting - Get troubleshooting help

Vector Database Endpoints

• GET /api/semantic-search?query=<search_term>&limit=<number> - Semantic search
• GET /api/products/enhanced?query=<search_term>&category=<category> - Enhanced product search

🧪 Testing

Run the test cases in test-cases.md to verify functionality:

1. Example Queries (Instalily Requirements):

   • "How can I install part number PS11752778?"
   • "Is this part compatible with my WDT780SAEM1 model?"
   • "The ice maker on my Whirlpool fridge is not working. How can I fix it?"

2. Feature Testing:

   • Basic chat functionality
   • Product search and compatibility checks
   • Installation guides and troubleshooting
   • Responsive design on mobile/desktop

🗄️ Vector Database Setup (Optional)

Pinecone Setup

1. Sign up at Pinecone
2. Create a new project and get your API key
3. Add the key to config.js
4. Restart the server - the system will automatically create the index

Fallback Mode

If no vector database is configured, the system automatically falls back to keyword-based search.

🔍 Usage Examples

User: "I need a water filter for my refrigerator"
AI: "I can help you find a water filter. What's your refrigerator model?"

User: "How can I install part number PS11752778?"
AI: [Provides detailed installation guide with steps and tools]

User: "Is this part compatible with my WDT780SAEM1 model?"
AI: [Shows compatibility results with product recommendations]

📖 Documentation

• Technical Documentation: See TECHNICAL_DOCS.md for detailed architecture and implementation
• Test Cases: See test-cases.md for comprehensive testing scenarios
• Project Summary: See PROJECT_SUMMARY.md for project overview and achievements

🤝 Contributing

This is a case study project by Bob Tianqi Wei for Instalily AI.

📄 License

This project is created for educational and demonstration purposes.

---

Developer: Bob Tianqi Wei
Project: Instalily AI Case Study
Date: June 2025