A company wants a customer-support application to answer questions using private company documentation. The company doesn't want to retrain its foundation model whenever the documentation changes.

Which approach is most appropriate?

An ML engineer needs to train, deploy, and monitor machine learning models on AWS.

Which service is most directly suited to these requirements?

A development team wants to build a generative AI application using managed foundation models without managing the underlying model infrastructure.

Which service should they consider?

An AI application needs to perform a multi-step task and decide which tools or external systems to use.

Which concept best describes this capability?

An organization wants its AI application to retrieve current internal information before generating an answer.

Which architecture is most appropriate?