Design an agent that handles refund requests end to end: it reads the customer's message, looks up the order, applies the refund policy, and either issues the refund, offers an alternative, or escalates.
The agent moves money. Everything about the design follows from that.
Build the architecture on a canvas: place the components, configure them, connect them into a data flow, and write a short reason for each one. The AI reviewer grades your design against a rubric written specifically for this problem.
Where does the refund policy live, and why there rather than in the prompt?
A customer sends a persuasive message arguing for a refund well over the limit. Trace what happens and say what stops it.
The refund call times out and the agent does not know whether it succeeded. What happens next, and what makes that safe?
Give every condition that sends a conversation to a human.
Minimum 8 components · needs a wide desktop screen