Smart Retail Assistant - RAG-Based Product Chatbot with .NET 9 and Azure SQL

[beneditotulio]

Project name

Smart Retail Assistant

Description

The Smart Retail Assistant is a high-performance RAG (Retrieval-Augmented Generation) chatbot designed to help users navigate a Walmart-like product catalog using natural language.

What problem is it solving?
Traditional retail search often fails when users have complex, intent-based queries (e.g., "I need a gift for a 10-year-old who likes space"). Our assistant solves this by leveraging semantic vector search to find relevant products that keyword-based search might miss.

What is it doing?

• Semantic Retrieval: It uses the native VECTOR data type and VECTOR_SEARCH in Azure SQL to find the most relevant products based on the user's intent.
• Grounded Responses: It uses .NET 9 and OpenAI to generate helpful responses grounded strictly in the retrieved SQL data, preventing AI hallucinations.
• Modern UX: Provides a clean, responsive chat interface with product recommendations, match percentages, and source attribution.

Architecture:

• Backend: ASP.NET Core 9 Web API
• Database: Azure SQL Database (with 2025 Vector Preview features)
• AI: OpenAI gpt-4o-mini and text-embedding-3-small
• Frontend: HTML5, Tailwind CSS, JavaScript

Type

RAG‑Based Chatbot Agent

Project Repository URL

https://github.com/beneditotulio/smart-retail-assistant

Project video (please verify that the link works)

https://youtu.be/58NQYUGr30s

[toolboc]

@beneditotulio, your video submission link is marked private, any chance you could flip it public or unlisted and update us?