Give your Agent memory with SQL Server and Microsoft Agent Framework | Data Exposed
Microsoft Developer explains how to add persistent memory to an AI agent using Microsoft Agent Framework, storing conversation history in SQL Server and showing how the same approach can be future-proofed with Azure SQL Database.
Overview
The episode focuses on making an agent “remember” users and prior conversations by plugging a database-backed memory store into Microsoft Agent Framework.
What the video covers
- A sample e-commerce application scenario
- Agent-driven recommendations
- A code walkthrough showing how agent memory is wired up
- How Microsoft Agent Framework supports memory via:
- A context provider
- A history provider
- Using the mssql-python driver from Python to connect to SQL Server / Azure SQL Database
- How the approach can move from local development to cloud with minimal changes
- Final tips, including notes around the Agent Framework UI
Technologies and resources referenced
- Microsoft Agent Framework (overview on Microsoft Learn)
- SQL Server in Docker (SQL Server on Linux containers)
- Azure SQL Database with Python using the mssql-python driver
Chapters (from the video description)
- 0:00 Intro
- 0:45 Sample e-commerce app
- 1:49 Agent recommendations
- 3:00 Agent code walkthrough
- 3:20 Context provider and history provider with Microsoft Agent Framework
- 5:04 More details on the history provider and how Microsoft Agent Framework helps
- 6:10 mssql-python driver
- 8:54 What else you can do with the framework
- 9:00 Local to cloud, easily pluggable
- 10:10 Final tips and trips, Agent Framework UI
Links (from the video description)
- Data Exposed playlist: https://aka.ms/dataexposedyt
- Microsoft Azure SQL YouTube: https://aka.ms/msazuresqlyt
- Microsoft SQL Server YouTube: https://aka.ms/mssqlserveryt
- Microsoft Developer YouTube: https://aka.ms/microsoftdeveloperyt
- AzureSQL Twitter: https://aka.ms/azuresqltw