It's silly to complain about LLMs being slow at tasks. Like sure, it's slower than you, but you can now spin up 10 agents on the 10 tasks and get the whole thing done quicker. You should be very happy to give up speed (on a per-task basis) for the benefit of autonomy.