AI agents can follow prompts and call tools, but they often lack the persistent organizational context needed to explain decisions, adapt over time, and maintain consistency across interactions.

In this session, you’ll learn how to build a knowledge layer for AI agents using Microsoft Foundry and Neo4j through context graphs. Instead of relying only on short-term conversation memory, context graphs model relationships, decision traces, historical interactions, and organizational knowledge in a structured and queryable way.

Through practical demonstrations, we’ll explore how to connect Neo4j with an AI agent stack in Microsoft Foundry to create agents that can retrieve relevant context, reason over prior interactions, and provide more explainable and auditable responses. You’ll see how graph-based memory enables persistent understanding, personalization, and context-aware decision-making without overwhelming the model with unnecessary information.

The session focuses on context engineering and the emerging role of knowledge layers as a foundation for scalable, explainable, and enterprise-ready AI agents.