PartSelect Chat Agent

PartSelect Chat Agent is an intelligent conversational system designed to assist users with appliance parts selection and repair guidance. The agent combines advanced RAG (Retrieval-Augmented Generation) technology with SQL database integration to provide accurate, context-aware responses for appliance maintenance and repair queries.

Key Features

• 🔍 Smart parts identification and pricing information
• 🛠️ Step-by-step repair guidance
• 📚 Access to comprehensive repair documentation and blog posts
• 🤖 Context-aware conversation handling
• 💡 Intelligent retrieval system combining RAG and SQL for accurate responses

System Architecture

This flowchart shows the system's architecture including:

• Database structure (Parts, Blog, Repair tables)
• Retrieval process using RAG + SQL
• Chat loop flow with query processing and response generation

Interface Demo

Installation

Prerequisites

• Python 3.11 or higher
• Conda or pip package manager
• Git

Setting up the environment

1. Clone the repository:

git clone https://github.com/zehuiwu/partselect-agent.git
cd priceselect-agent

2. Create and activate a new conda environment:

conda create -n new_env python=3.11
conda activate new_env

3. Install the required packages:

pip install -r requirements.txt

Environment Variables

Create a .env file in the root directory with the following variables:

OPENAI_API_KEY=your_openai_api_key
DEEPSEEK_API_KEY=your_deepseek_api_key

Replace your_*_api_key with your actual API keys.

Note: Make sure to keep your .env file private and never commit it to version control.

MySQL MCP Server Setup

1. Install MySQL:

   • For macOS: brew install mysql
   • For Ubuntu: sudo apt-get install mysql-server
   • For Windows: Download and install MySQL from the official website

2. Start MySQL service:

   • For macOS: brew services start mysql
   • For Ubuntu: sudo service mysql start
   • For Windows: MySQL service should start automatically

3. Import the data:

   # Create the database
   mysql -u root -p -e "CREATE DATABASE partselect;"
   
   # Import the CSV files
   mysql -u root -p partselect < data/all_parts.csv
   mysql -u root -p partselect < data/all_repairs.csv

4. Add the following environment variables to your .env file:

   MYSQL_HOST=localhost
   MYSQL_PORT=3306
   MYSQL_USER=root
   MYSQL_PASSWORD=your_password
   MYSQL_DATABASE=partselect
   

   Replace your_password with your MySQL root password.

Note: Make sure the MySQL server is running before starting the PriceSelect Chat Agent.

Starting the Web Interface

To start the web interface:

1. Navigate to the web client directory:

   cd lily_client/web

2. Start the server with hot reloading enabled:

   hypercorn app:app --reload

3. Open your web browser and navigate to http://localhost:8000 to access the web interface.