Large language models are powerful—but by default, they forget everything the moment a conversation ends. In real business scenarios, that’s not enough.
In this session, you’ll learn how to design stateful Copilot agents that can selectively remember what matters and ignore what doesn’t—without custom code or external databases.
We’ll walk through a real-world pattern using Copilot Studio, SharePoint Lists, and Power Platform connectors to implement persistent agent memory. You’ll see how an agent can:

Decide when information is worth remembering
Store that memory in a governed, auditable system (SharePoint)
Reuse memories as a knowledge source in future conversations
Avoid becoming a “data hoarder” by filtering low‑value inputs

Along the way, we’ll cover design considerations like memory scoping, trust boundaries, data hygiene, and why “remember everything” is usually the wrong answer.
You’ll leave with a practical architecture you can apply immediately—whether you’re building internal copilots, customer-facing agents, or reusable enterprise patterns.
Ideal for: Copilot Studio makers, Power Platform professionals, solution architects, and anyone building AI agents that need context beyond a single chat.

*Other Title options:
- Designing Stateful Copilot Agents: Persistent Memory with Copilot Studio and SharePoint
- Building Stateful Copilot Agents: Implementing Selective Memory with Copilot Studio
- I’ll Remember That”: Teaching Copilot Studio Agents What Actually Matters
*Subtitle:
-Implementing governed, persistent memory using SharePoint and Power Platform
*Tagline:
Because the most useful Copilot isn’t the one that knows everything—it’s the one that remembers the right things.
*Learning Objectives:
1. Explain the difference between stateless and stateful Copilot agents
Understand why default Copilot interactions forget context, and when persistent memory is necessary for real business scenarios.
2. Design a selective memory pattern using Copilot Studio and SharePoint
Learn how to let an agent decide what information is worth remembering, store it in a governed SharePoint List, and reuse it as a knowledge source.
3. Apply governance and quality controls to agent memory
Identify best practices for memory scope, data hygiene, and avoiding low‑value or risky information from being stored long‑term.