Improving search relevance typically requires complex personalization pipelines — recommendation engines, feature stores, ML models. This session shows a simpler path: multiple lightweight agents that collaborate through OpenSearch's agentic memory to understand and enrich queries in real time. Same query, different results for different users.