Question 37
Domain 1: Design ApplicationsAn internal HR assistant must answer questions from policy documents that change every week and cite the source material. Which approach is the best starting point?
Correct answer: B
Explanation
A RAG application is the best starting point because it retrieves answers from the latest policy documents instead of relying on a fixed model memory. This fits the need to answer from documents that “change every week” and to cite the source material, since retrieval can return the exact policy text used for each response.
Why each option is right or wrong
A. Fine-tune a chat model on past HR tickets
Fine-tuning bakes patterns into model behavior, but weekly policy updates are better handled through retrieval.
B. Build a RAG application over the policy documents
A retrieval-augmented generation setup is the appropriate first implementation when the knowledge base is volatile, because it grounds each answer in the current document set rather than in model parameters that are only updated at training time. In practice, the retrieval layer can surface the exact policy passages used to answer the question, which supports source citation and avoids the staleness problem that would arise if the assistant relied on a static fine-tune or prompt-only approach.
C. Train a sentiment classifier for each policy area
Sentiment classification labels tone or opinion, not policy question answering with grounded citations.
D. Hardcode the answers into the system prompt
Hardcoded prompts become stale quickly and do not scale for frequently changing policy content.