Inside OpenSearch’s bid to become the default AI data layer

OpenSearch 3.5 and 3.6 add dense and sparse vector retrieval, hybrid search, and binary quantization to better support production AI workloads.

The releases also introduce native agent memory, letting conversational apps store and retrieve context directly in OpenSearch.

With ML Commons, Lucene, Faiss, and HNSW enhancements, OpenSearch is moving closer to a default AI application data layer.

Teams can consolidate semantic search and session memory on existing OpenSearch infrastructure instead of adding separate vector databases or stores.