Question

A development team is building a RAG-based application on AWS. They have already selected a foundation model and chunked their internal documents. To enable semantic retrieval, they need a component that stores the numerical representations of the document chunks and supports fast similarity search at query time. Which component fulfills this requirement?

Accepted Answer

A vector database (vector store) that indexes embeddings of the document chunks

Answer

A relational database with the raw document text stored in indexed columns

Answer

An object storage bucket holding the original PDF and Word files

Answer

A message queue that buffers user prompts before sending them to the model