Understanding Agent Memory in Microsoft Foundry | Step-by-Step Demo

In this video, I explain Agent Memory in Microsoft Foundry in a clear and practical way, and then demonstrate it using a real Python lab.

Most AI agents today are stateless. They understand you during a conversation, but the moment the session ends, they forget everything. That is fine for demos, but it breaks real-world applications.

Microsoft Foundry Agent Memory solves this problem by providing managed, long-term memory for agents. It allows agents to remember user preferences, constraints, and past interactions across conversations, without you building custom databases or memory pipelines.

In this video, you will learn:

What Agent Memory is in Microsoft Foundry
How memory is different from chat history
Memory types: user profile memory and chat summary memory
What a memory store is and why you need one per agent
Why memory scope is mandatory and how it ensures user isolation
How memory is extracted, consolidated, and retrieved
Security, privacy, and deletion considerations
Limits and pricing basics

After the explanation, I walk through a complete hands-on lab where I build a Trail Finder Agent that remembers walking and hiking preferences such as distance, incline, terrain, and fitness level.
You will see memory being created, recalled across conversations, updated over time, isolated per user, and deleted when required.

Github - https://github.com/Shailender-Youtube/Microsoft-Foundry-Agent-Service-Memory

Видео Understanding Agent Memory in Microsoft Foundry | Step-by-Step Demo канала MadeForCloud