Some scenarios have no dependencies on mainstream processes, and you can use tasks directly to implement similar features. The framework provides both a co-process and an asynchronous task for the developer. Keep in mind that whether it's a co-process task or an asynchronous task, the tasks inside the task support only the collaboration, and only all IO operations encapsulated by the framework (database, cache ...) are used.
When a cointegration task delivers a task, it does not block the main process from being the equivalent of a federation call, typically for scenes that need to wait for the result of the task to return.
Some scene main processes do not care about the results of task execution, at which point you can use asynchronous tasks.