A retailer refunds an online order when the amount is under a published threshold and the item is unused. That path follows the rule and does not ask a model to judge the shopper. Which component should run the refund?
Select an answer to reveal the explanation.
Short Explanation
A clerk with the refund rule on a card does not stop to invent a new policy. You want an agent flow that posts the refund when the amount is under the threshold and the item is unused. Asking a model who deserves the money adds a judgment this path does not have.
Full Explanation
An agent flow is the component for a path whose branches are already decided and whose work is a fixed sequence, such as reading the amount, checking that the item is unused, and posting the refund. The retailer has published that rule, and the requirement says the path does not ask a model to judge the shopper, so the flow is the design. A prompt action that asks a model whether this particular shopper deserves the money back replaces the published rule with a fresh judgment on every order. A fallback topic runs when a conversation matches nothing else, and it is not the component that posts a refund against an order record. A custom Foundry model trained on past refunds would score who should receive one, which is a prediction the business did not ask for and a poor substitute for a threshold. Exam caveat: a threshold can still be a business rule when people might argue the edge cases. This design has closed those cases. Operational check: submit one unused order under the threshold and confirm the flow posts the refund with no model step, then submit one over the threshold and confirm no refund is posted.