Give your Agent memory with SQL Server and Microsoft Agent Framework | Data Exposed

A walkthrough of building an AI agent with persistent conversation memory using Microsoft Agent Framework and SQL Server. The demo shows an e-commerce recommendation system where user preferences (stored per session in SQL Server via Docker) are retrieved by the agent to personalize product suggestions. The Microsoft Agent Framework is explained as a merger of AutoGen and Semantic Kernel, supporting multi-agent orchestration, Azure integrations, and a built-in Dev UI for local debugging and tracing. The history provider pattern allows swapping SQL Server for any other data source, and the local setup can be migrated to Azure with a connection string change.