AWS OpenSearch General Manager to Outline Retrieval Strategy for AI Agents at OpenSearchCon

AWS OpenSearch General Manager Tia White is scheduled to deliver a keynote at OpenSearchCon North America, running September 22-24 in San Jose, where she will discuss how AWS is combining full-text search, vector retrieval, hybrid search and operational analytics to address the context requirements of enterprise AI agents and the unpredictable compute costs they create.

White said hybrid search is especially relevant to enterprise AI: semantic retrieval can find information by meaning, while lexical matching preserves precision for product identifiers, account numbers and technical terminology. AWS sees an opportunity for customers to extend existing OpenSearch deployments into AI workloads, potentially avoiding the need for another database, security setup and operational workflow.

To address capacity planning challenges caused by agents that repeatedly search, refine queries and retrieve again, AWS made OpenSearch Serverless generally available in May 2026. The service separates compute and storage, with resources provisioned in seconds, autoscaling up to 20 times faster than its predecessor, and claimed savings of up to 60% versus provisioning for peak demand, according to AWS.

The company also highlighted its optimized log analytics engine, which combines columnar analytics with full-text search and is advertised at up to four times better price-performance and twice the analytical-query speed, helping teams investigate agent failures. OpenSearch 3.5 introduced persistent agent conversation memory, and Amazon OpenSearch Service announced support for the version in March 2026.