Seems like you're dismissing the cost of artificially splitting what should be a monolith. It outweighs the supposed productivity boost of LLMs by at least one

Seems like you're dismissing the cost of artificially splitting what should be a monolith. It outweighs the supposed productivity boost of LLMs by at least one order of magnitude if not more.