Single-shot and agentic retrieval with Amazon Bedrock Managed Knowledge Base
Amazon Bedrock Managed Knowledge Base now supports single-shot RAG and retrieval-only requests through AgenticRetrieveStream, alongside agentic retrieval. This post explains how the maxAgentIteration parameter matches retrieval effort to each request, using an illustrative internal banking policy assistant.
Amazon Bedrock Managed Knowledge Base now offers single-shot Retrieval Augmented Generation (RAG) and retrieval-only requests via AgenticRetrieveStream. With single-shot RAG, the service conducts a single search using a foundation model to generate an answer from the retrieved passages. Retrieval-only requests return the matching passages without generating an answer, allowing the application or user to work with the source text directly.
These modes complement the existing agentic retrieval, which plans and executes multiple searches before answering. The same API can now handle both the employee's simple and complex questions. Single-shot RAG is suitable for direct questions like checking a checklist, while agentic retrieval excels at comparing policies across different business structures.
Agentic retrieval uses a foundation model to break down the question into sub-queries, search relevant knowledge bases, and evaluate the results for answer generation. Developers can benefit from better support for multi-part questions, targeted searches, and answers with supporting evidence and citations. The range for maxAgentIteration is now 0-5, offering flexibility to balance speed and answer quality based on the request's complexity.
Written by urgent.news from AWS Machine Learning's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
