Turning Agent Memory Into Skills That Work — Will Lyon, Neo4j

Agents finish a task, then forget what they learned. Retrieval alone won't fix that. Will Lyon, senior product manager at Neo4j, explains why agent memory built on embeddings and retrieval isn't enough: recall isn't actionable knowledge. Agents need knowledge that is connected, typed and traceable. He shows Neo4j's model of agent memory as a context graph with short-term, long-term and reasoning memory, covering entity extraction and resolution, shared ontologies, and decision traces that let many agents learn from each other. Then he turns memory into action: distilling grounded, governed skills from the graph, based on Neo4j research that treats skills as typed execution graphs, with checks for grounding, coverage and coherence, and alerts when a skill goes stale. He demos the Neo4j Agent Memory Service distilling a patient-intake skill from a healthcare conversation. In this talk: • Why retrieval isn't actionable knowledge • Short-term, long-term and reasoning memory in one context graph • Distilling grounded, governed skills from memory • Demo: the Neo4j Agent Memory Service building a skill from a conversation SPEAKER William Lyon, Senior Product Manager, Neo4j LinkedIn: https://www.linkedin.com/in/lyonwj X: https://x.com/lyonwj Website: https://lyonwj.com LINKS Neo4j Agent Memory Service (NAMS): https://memory.neo4jlabs.com Neo4j on X: https://x.com/neo4j CHAPTERS 0:00 Intro 0:42 The amnesiac loop 1:27 Recall isn't actionable knowledge 2:06 Short-term, long-term and reasoning memory 2:36 From messages to a knowledge graph 3:06 Entity resolution and ontology 4:01 The reasoning graph 4:41 Decision traces 5:26 Shared memory across agents 6:01 From memory to skills 6:55 Agent skills, explained 7:50 The problem with prose skills 8:30 Research: skills as graphs 9:35 Benchmark results 10:05 Distilling skills from memory 10:50 Grounding, coverage and coherence 11:39 Skill governance 12:14 The Neo4j Agent Memory Service 13:04 The distillation pipeline 14:14 Composing skills 14:39 Demo: the NAMS dashboard 15:58 Demo: distilling a skill 16:53 Demo: a patient-intake skill 17:53 Resources Recorded at the AI Engineer World's Fair 2026 in San Francisco. Subscribe for more talks from the engineers building with AI. AI Engineer: https://ai.engineer YouTube: https://www.youtube.com/@aiDotEngineer X: https://x.com/aiDotEngineer LinkedIn: https://www.linkedin.com/company/aidotengineer/ #KnowledgeGraph #AgentMemory #AIEngineer