//
Busque em todas as páginas da documentação
8 pages in this section.
Ingestion, chunking, embedding, and retrieval as one managed pipeline - and the two client families (bedrock-agent, bedrock-agent-runtime) that configure and query it.
Create a Knowledge Base over an S3 data source via SDK - control-plane setup, ingestion, and your first Retrieve and RetrieveAndGenerate calls in boto3 and AWS SDK v3.
Multi-format ingestion (PDF, HTML, tables) without a custom parser - configuring chunking strategy and Smart Parsing on a Data Source in boto3 and AWS SDK v3.
Managed RAG orchestration vs building your own retrieval-augmentation loop - RetrieveAndGenerate compared against Retrieve plus a hand-built Converse prompt, with code for both.
A checklist for running Amazon Bedrock Knowledge Bases from the SDK safely and cheaply - chunking strategy, ingestion hygiene, retrieval-quality evaluation, cost, and IAM security.