I am describing the reality we are currently in. If I don't do some harness engineering then my bots crap on the floor and I start questioning whether I should delegate to them at all and if me doing it manually wouldn't still take less time.
And you are describing a desired reality. I sympathize, mind you, it's just not the one we are currently living in.
Yes, you can vibe up a demo in no time. But LLMs still need guidance to produce an architecture that will hold up to real world scenarios.