Question 3
Section 3A customer-service team wants to reduce average response time by drafting suggested replies for agents, while agents still approve the final message. Which generative AI pattern best fits this use case?
Correct answer: C
Explanation
Human-in-the-loop response assistance fits because the AI drafts suggested replies, but a person still reviews and sends the final message. This matches the pattern where generative AI supports work while “agents still approve the final message,” keeping human oversight in the workflow.
Why each option is right or wrong
A. Replacing all agents with an unsupervised system
Full replacement removes human approval, which contradicts the stated need for agent review.
B. Training a custom chip for model serving
Custom chip training concerns infrastructure, not the workflow pattern for assisted customer responses.
C. Human-in-the-loop response assistance
The workflow keeps a person in the approval step, so the AI is only assisting with drafting rather than autonomously handling customer communication. In generative AI pattern terms, that is the human-in-the-loop model: the system proposes content, and the human retains final control over what is sent, which is exactly what the prompt describes.
D. Disabling historical support data because it contains examples
Historical support examples are generally useful training or grounding data, not something to disable.