This video demonstrates how to use the new chat with your document feature in Knowledge Bases for Amazon Bedrock. It explains how retrieval-augmented generation (RAG) allows you to bring your own data to generative AI without setting up complex knowledge bases. The presenter shows how to use this feature through the AWS Console and the RetrieveAndGenerate API in the Python SDK. The demo uses a document about 1970s music to showcase asking questions and getting relevant responses. This feature currently supports Claude 3 Sonnet model and various document formats up to 10MB or 20,000 tokens.