Amazon Web Services Inc. today announced the general availability of vector search to DynamoDB, the company’s high-availability NoSQL key-value and document database designed for high speed and scale.

Launched in 2012, the database service has gone through numerous iterations and is positioned today as the go-to database for developers who need predictable performance without operational overhead. It is a powerful service for serverless web and mobile apps, gaming, ad tech, internet of things, retail and applications that require low-latency access at high throughput.

Today’s announcement brings vector search with single-digit millisecond latency and 99% recall capability. It is designed to support any scale, including trillions of vectors. Developers do not need to provision, patch or manage any servers.

This update will allow the database to support large-scale applications that require semantic retrieval, including artificial intelligence applications and agentic memory, retrieval-augmented generation, recommendation engines, personalization and anomaly detection.

Vector database capabilities are critical to AI development because they store high-dimensional data called embeddings, which allow for fast similarity searches based on meaning rather than keyword matching. Many AI systems use vector databases as long-term memory and to handle large decks of unstructured data. These systems help build context for large language models and agents to increase response accuracy and prevent false outputs, such as hallucinations.

The benefit of adding this capability to DynamoDB is that numerous developers already use the service for retrieval of other data. This means that developers can now move their vector search capabilities onto a service that also serves their operational data, sharing the same managed serverless infrastructure and the same pay-per-request pricing.

Before this update, developers would have needed to run their primary data scale storage and vector search separately.

The company currently hosts numerous native vector search capabilities across numerous specialized databases; the vector primary services include S3 Vectors, a native storage, indexing, and sub-second similarity query engine for cloud object storage and OpenSearch Service, which offers fully managed vector search with a serverless vector engine for billion-scale datasets.

Image: SiliconANGLE/Microsoft Designer