Question 3
Domain 5: Security, Compliance, and Governance for AI SolutionsAn airline company wants to build a conversational AI assistant to answer customer questions about flight schedules, booking, and payments. The company wants to use large language models (LLMs) and a knowledge base to create a text-based chatbot interface. Which solution will meet these requirements with the LEAST development effort?
Correct answer: B
Explanation
Amazon Bedrock Knowledge Bases manages the full RAG pipeline: “ingesting documents,” “chunking them,” “generating vector embeddings,” and “retrieving relevant chunks at inference time.” That lets the airline use LLMs plus a knowledge base in a text chatbot with minimal custom orchestration, and “Agents for Amazon Bedrock” can query Knowledge Bases and external APIs in one session.
Why each option is right or wrong
A. Train models on Amazon SageMaker Autopilot.
SageMaker Autopilot automates model training, not a knowledge-base chatbot workflow.
B. Develop a Retrieval Augmented Generation (RAG) agent by using Amazon Bedrock.
Amazon Bedrock Knowledge Bases is the managed RAG feature that ingests source content, chunks it, creates embeddings, and retrieves relevant passages at inference time, so the airline does not need to build its own vectorization or retrieval pipeline. For a text chatbot over schedules, booking, and payment FAQs, this is the lowest-effort path because it uses a foundation model plus a knowledge base in one managed service rather than custom training or hand-rolled orchestration. Agents for Amazon Bedrock can then invoke the knowledge base and external APIs in the same session if the assistant must also perform actions like booking lookups or payment-related workflows.
C. Create a Python application by using Amazon Q Developer.
Amazon Q Developer is for developer assistance, not building customer-facing chatbot applications.
D. Fine-tune models on Amazon SageMaker Jumpstart.
SageMaker JumpStart accelerates model deployment and fine-tuning, not turnkey RAG chatbot creation.