Make Your AI Agents Predictable With Graph Engineering - Zachary Blumenfeld, Neo4j

Graph engineering is a new name for an old idea: give agents a map of the logic they should follow, not just a pile of text. Zachary Blumenfeld of Neo4j explains how graphs make agents more predictable, governable and portable. He shows how to compile skills and past decision traces into graph-shaped skills, then demos a loan-approval agent that follows them. 0:00 Intro: what is graph engineering? 1:42 Three types of graphs 3:04 Predictability, governance, portability 4:42 Neo4j's knowledge layer 6:03 Context graphs and agent memory 6:53 Compiling skills into graphs 8:08 Distilling skills from memory graphs 9:22 Customer use cases 10:47 Demo: a loan-approval agent 14:30 Compiling runs into a graph-shaped skill 18:20 Observability on graph branches 20:01 Resources Zachary Blumenfeld, Neo4j: https://neo4j.com Join us in SF for the second annual AIE CODE summit: https://aie.codes our last flagship summit of 2026. Applications live now!