The LLM as Invisible Operator

Generative AI didn’t wait for the grand digital transformation roadmaps. It slipped in through the cracks: into executive assistants’ inboxes, into the Excel macros of financial controllers, into the Python scripts of engineers who had no time to submit a resource request.

The Operator With No Job Description

What’s striking is the absence of any nomenclature. Nobody says “the LLM produced this report.” They say “I produced this report.” The tool dissolves into the action. This is precisely what Ethan Mollick describes with the centaur: an augmented human who can no longer clearly separate their own work from the machine’s.

The phenomenon creates a ghost layer in the org chart. Reporting lines haven’t moved. Job titles haven’t changed. But the workflows have been quietly rewritten.

What It Actually Changes

First, the speed of first-draft production explodes. A document that took two days now takes two hours. That’s not productivity in the classical sense — it’s a redefinition of what “finished work” even means.

Second, the entry threshold for complex tasks drops sharply. A solo founder can now produce a competitive analysis memo that previously would have required a team of four.

Third — and this is where it gets interesting — value migrates toward judgment. Producing is cheap. Validating, contextualizing, deciding: that’s still human territory.

The Question That Remains

If the invisible operator does its job well, who owns the outcome? The legal and organizational answer is still murky. That’s the real construction site for the next three years.