Next-Generation OpenSearch: With a Side of Agentic AI

Search technology has evolved from keyword-matching-focused lexical search to semantic search, which understands meaning, then to hybrid search, which combines both approaches, and now to the stage of Agentic Search, where AI agents reason on their own and perform multi-step searches.

Amazon OpenSearch Service has introduced the following key features to improve operational efficiency.

• Cluster Insights: Automatically identifies bottlenecks by visualizing node performance and query patterns.
• Derived Source: Dynamically reconstructs the original from indexed fields, cutting storage costs by up to 40% and improving indexing speed by 20%.

It also offers a tiered storage structure to economically manage large-scale vector data of over 1 trillion items. You can choose the optimal option based on your data characteristics, ranging from Exact KNN to In-Memory, Disk Mode (utilizing quantization), and the most affordable S3 vector storage. Notably, through collaboration with NVIDIA, it also supports GPU index acceleration based on cuVS.