Road to NODES 2026: Agents in the Wild: Building AI That Learns
Play video on YouTube: Road to NODES 2026: Agents in the Wild: Building AI That Learns
Up next
Road to NODES AI: Graph-Based Long-Term Memory: How Agentic Workflows Adapt Through Experience
Up next
Your agent works great in the demo, then forgets everything the moment the task changes. Hard-coded workflows can’t learn, and fully autonomous agents are hard to trace.
In this workshop, we build agent memory as a graph in Neo4j. Using value investing as the test case (ownership structures, peer comparisons, financial ratios), you will:
Compare a predefined workflow with an autonomous agent and see how their execution traces differ
Model agent traces as a graph and query them with Cypher to find lessons learned
Build hierarchical context graphs and compose playbooks at query time
Close the Reflector→Curator loop with human feedback and real-world outcomes
You also get an honest look at where graph memory pays off and what it costs in latency and complexity.
Trainer: Alessandro Negro
Resources mentioned





