Turn Your Database into a Smart Chatbot with Azure OpenAI, LangChain, and ChromaDB

In recent years, language models and AI have reshaped how we interact with data. Yet, a significant portion of business knowledge still resides in relational databases (highly structured but not exactly user-friendly for those unfamiliar with SQL). So what if you could query that data using natural language, as if you were speaking to a smart assistant?

In this post, we’ll walk through a practical proof of concept (POC) using the well-known Northwind database as our data source. We’ll apply Retrieval-Augmented Generation (RAG) techniques to convert structured data into meaningful, searchable knowledge. By combining tools like LangChain, OpenAI Embeddings, and ChromaDB, we’ll build a system that can answer real business questions about customers, orders, and sales (all without writing a single SQL query).

This hands-on example will demonstrate how to turn raw database records into descriptive text, generate semantic embeddings, and store them in a vector database ...