A Virtual Sales Agent that uses LangChain, LangGraph, and Gemini Flash to simulate customer interactions. Features include product inquiries, order management, and personalized recommendations through a user-friendly Streamlit interface.

Virtual Sales Agent powered by LangGraph, Streamlit & Gemini Flash

This project implements a Virtual Sales Agent that simulates customer interactions, providing information and support through a Streamlit interface. Using the power of LangChain, LangGraph, and a SQLite database, this agent can answer product questions, create orders, check order statuses, and offer personalized recommendations. These tools are divided into safe and sensitive categories. For sensitive tools, such as creating orders, a human-in-the-loop mechanism is implemented, requiring approval or denial before proceeding.

👉 Check out a quick demo of the Virtual Sales Agent in action in the Interface Preview section!

Table of Contents

1. Key Features
2. Built With
3. Use Cases
4. Project Structure
5. Interface Preview
6. Get Started
7. Contributing
8. Related Resources
9. Future Plans
10. License

Key Features

This virtual sales agent can assist customers with:

1. Product Inquiries:

   • Answer questions about product availability, pricing, and stock levels.
   • Example Questions:
     • "What products do you have in stock?"
     • "How much does product X cost?"
     • "Is product Y available?"

2. Order Placement:

   • Allow customers to create new orders, referencing data from the database.
   • Example Request: "I would like to order 2 units of product Z."

3. Order Tracking:

   • Provide up-to-date status information for existing orders.
   • Example Question: "What is the status of order #54321?"

4. Personalized Recommendations:

   • Suggest relevant products based on a customer's past purchase history.
   • Example Recommendation: "Based on your previous order, you might also