AWS Certified Generative AI Developer - Professional AIP-C01

A company operates a Retrieval Augmented Generation (RAG) application that uses Amazon Bedrock Knowledge Bases for regulatory compliance queries. The application uses the RetrieveAndGenerateStream API. It retrieves relevant documents from a knowledge base containing more than 50,000 regulatory documents, legal precedents, and policy updates.

The RAG application is generating suboptimal responses because initial retrieval frequently returns documents that are semantically similar but contextually irrelevant. These poor responses are leading to model hallucinations and incorrect regulatory guidance. The company must improve RAG application performance so that it returns more relevant documents.

Which solution meets this requirement with the LEAST operational overhead?