In n8n, I use an AI agent node. Assume that this agent node is the main agent, and there are several sub-agents under it to complete different tasks. Then I want to ask, if there is a task that requires different agents to collaborate and execute in a certain order, do I need to write the order of all sub-agents that need to execute the task in the main agent, or do I not need to write it in the main agent, as long as the task enters the main agent, it will automatically determine how to call the sub-agents and in what order to call them?