Amazon DynamoDB now supports native vector search, giving developers a way to store embeddings alongside application data and run similarity queries directly inside the managed NoSQL database.
The feature adds a vector index type and a SearchVectors API. Developers can use embeddings from models such as Amazon Bedrock Titan Text Embeddings, Cohere Embed, or OpenAI text embedding models, then configure dimensions and distance functions for approximate nearest-neighbor search.
For retrieval-augmented generation, recommendations, agent memory, and semantic search, the change removes a common data pipeline. Teams no longer have to copy application data into a separate vector database and keep both systems synchronized just to support similarity lookup.
AWS says the feature supports up to 4,096 dimensions, Euclidean, cosine, and dot-product distance functions, plus inline filtering. Costs are separate from normal table charges and are metered for data written to the index, data processed during search, and data stored. That makes index design and projection choices important for production use.