AWS announced updates to Amazon Bedrock knowledge bases, a new capability announced at AWS re:Invent 2023 that allows organizations to provide information from their own private data sources to improve the relevance of answers.
According to AWS, there have been significant improvements since launch, such as the introduction of Amazon Aurora PostgreSQL-Compatible Edition as an additional option for custom vector storage alongside other options such as the vector engine for Amazon OpenSearch Serverless, Pinecone and Redis Enterprise Cloud.
One of the new updates Amazon is announcing is expanding the choice of built-in models. With Amazon Titan Text Embeddings, users can now choose between Cohere Embed English and Cohere Embed Multilingual models, both of which support 1024 dimensions, to convert data into vector embeddings that capture the semantic or contextual meaning of text data. This update aims to give customers more flexibility and precision in how they manage and use their data within Amazon Bedrock.
To offer greater flexibility and control, knowledge bases support a choice of custom vector stores. Users can choose from a range of supported options, customizing the background to their specific requirements. This customization extends to providing vector database index names, along with detailed mappings for index fields and metadata fields. Such features ensure seamless and efficient integration of knowledge bases with existing data management systems, increasing the overall utility of the service.
In this latest update, Amazon Aurora PostgreSQL compatible and Pinecone serverless have been added as additional choices for vector stores.
Many of the Amazon Aurora database features will also be applied to vector embedding workloads, such as elastic storage scaling, low global read latency, and faster throughput compared to open source PostgreSQL. Pinecone serverless is a new serverless version of Pinecone, which is a vector database for building generative AI applications.
These new capabilities provide users with greater variety and scalability in their choice of vector storage solutions, enabling more customized and efficient data management strategies.
And finally, an important update to the existing Amazon OpenSearch Serverless integration has been implemented, aimed at reducing costs for users involved in development and testing workloads. Redundant replicas are now disabled by default, which Amazon estimates will cut costs in half.
Together, these updates underscore Amazon Bedrock’s commitment to improving the user experience and offering versatile, cost-effective vector data management solutions in the cloud, according to Antje Barth, chief developer advocate at AWS in blog post.