The method involves structuring the application by separating guardrails to clarify the agents’ thought processes.