The evolution of AI systems is shifting from simple retrieval to autonomous reasoning. This talk explores how knowledge graphs serve as the foundational layer for building intelligent, agentic AI systems that can plan, reason, and execute complex multi-step workflows in production environments.
I show how to use GraphRAG to unlock semantically aware retrieval through entities, relationships, and provenance, the real breakthrough comes when we layer reasoning models and agentic frameworks on top of this structured foundation. I will demonstrate how knowledge graph driven architectures enable LLM agents to move beyond static pipelines toward dynamic problem-solving whilst planning multi-step tasks, orchestrating tool usage, and maintaining explainable decision paths.
This talk goes beyond theoretical patterns to share battle-tested approaches for taking agentic systems to production. You will learn architectural patterns that bridge research prototypes and enterprise deployments, including strategies for managing agent state, ensuring reliability, monitoring agentic behavior, and building systems that scale.
Speaker
Cassie Shum
Vice President of Ecosystem, Product Engineering @RelationalAI, Previously @Thoughtworks
Cassie Shum is Vice President of Ecosystem, Product Engineering at RelationalAI, where she leads teams in designing and deploying production AI systems built on knowledge graphs. Previously at Thoughtworks, she worked across industries to architect data platforms and AI solutions that bridge cutting-edge research with enterprise realities. Cassie specializes in GraphRAG, agentic workflows, and building AI systems that are robust, explainable, and production-ready.
