This session covers how to build and maintain knowledge graphs that agents can query, update, and reason over. We'll cover graph schema design for agent memory, integrating Neo4j or other graph databases with LLM workflows, and techniques for agents to extract entities and relationships from conversations.

I'll also cover how to implement graph-based retrieval that understands context and connections, enables agents to perform multi-hop reasoning across related concepts, and maintains graph consistency as agents learn new information.

The session includes a live demo of an agent that builds its knowledge graph in real-time, plus practical patterns for when graphs outperform vector search and when to use hybrid approaches.