I think you misread. The point I make is that it's many agents = 1 task.
Since the probability of a LLM succeeding at any given task is sub 100%, you should run multiple of the same LLM with the same prompted task in parallel.
I think OP means they should be collaborating. In the posters proposed solution each agent is independent. But you could reduce the human attention required by having multiple rounds of evaluation and feedback from other agents before it gets to the human.
Since the probability of a LLM succeeding at any given task is sub 100%, you should run multiple of the same LLM with the same prompted task in parallel.