Building a Conversational MySQL Database System Using Langchain and Google Palm

This demo will showcase a system that allows natural language interaction with a MySQL database using Langchain and Google Palm. The system is designed for TrendVibe Tees, a T-shirt store, and enables store managers to ask questions like “How many white Adidas t-shirts are left in stock?” or “What is the projected revenue if we sell all extra-small t-shirts with discounts?”.

By converting natural language queries into SQL, the system retrieves accurate data from the MySQL database. The project integrates Hugging Face embeddings, Chromadb as a vector store, and Streamlit for the UI. It also uses few-shot learning to enhance query accuracy over time.

TrendVibe Tees Here is a fictional store, I will use it to demonstrate the idea, but the system can be applied to any business.

The project is still in progress, and I will share the project URLs (Github, and demo links) after the demo day.