Every few months a story does the rounds about a job that gets “optimised” out of existence. A door greeter, a receptionist, an admin. The pitch is always the same: we can automate the visible part of the role, save some money, and nothing important will change.
Except something important always changes.
Rory Sutherland tells the classic version with the luxury hotel doorman.
A consultant notices the doorman costs $40k a year, recommends an automatic door, and declares a cost saving victory. Two years later the hotel is a mess because the doorman wasn’t just opening the door, he was providing security, hailing cabs, making guests feel welcome, giving the place status.
He was a bundle of value disguised as a simple task.
That’s the trap. We judge things based on their Surface Value and miss the Real Value that sits underneath.
Surface Value is the part that’s easy to measure—door opens, email sent, report delivered. Real Value is everything else—trust, judgement, confidence, connection, capability. The stuff that makes the system work. When you optimise purely for the visible task, you often break the invisible ones. And in most organisations, the invisible ones are carrying the load.
I see this constantly in digital delivery. Someone looks at a coordinator and says, “All they do is upload models and send reminders.” So they buy a workflow tool. Then they wonder why the team’s still misaligned, the model’s still a mess, and the client’s still unhappy.
The Surface Value was the button‑pressing. The Real Value was the judgement, the relationships, the conflict‑management, the pattern‑spotting, the early warnings.
You can automate the clicks. You can’t automate the care.
Now layer AI on top of all this and the fallacy becomes even easier to fall for.
We’re in a moment where AI lets people produce pages of polished output without any of the underlying competence. People who’ve never delivered a project or managed a team are suddenly publishing “thought leadership” that looks convincing because the sentences are clean. It’s competence‑shaped content, not competence.
That’s the modern Doorman Fallacy. Assuming the artefact is the work.
- If you think writing is just getting words onto a page, you’ll outsource it.
- If you think modelling is just geometry, you’ll script it.
- If you think project delivery is just meetings, you’ll automate it.
The risk isn’t that AI exists. The risk is that people stop recognising the difference between output and understanding. And the real value—the human layer—usually hides under the surface:
- The side‑conversation in a meeting that dodges a blow‑up later.
- The quiet doubt that tells you a drawing “looks right but isn’t”.
- The years of pattern recognition that let you read a client’s tone before you read their words.
- The unspoken coordination between disciplines that makes a deadline possible.
The side‑conversation in a meeting that dodges a blow‑up later.
The quiet doubt that tells you a drawing “looks right but isn’t”.
The years of pattern recognition that let you read a client’s tone before you read their words.
The unspoken coordination between disciplines that makes a deadline possible.
None of that appears in the task list or the automation pitch. AI can assist it, speed it up, support it. But it can’t replace it.
I’m pro‑AI. But only when it protects the real value, not just the visible value. AI is great at clearing out the low‑level grind. It’s terrible at replacing judgement. If you let it handle the repetitive noise so you can focus on the real work—that’s a win. If you let it stand in for capability, you’re repeating the hotel’s mistake with a shinier automatic door.
If we’re not careful, AI won’t hollow out jobs. It’ll hollow out people, leaving behind professionals who look productive on the surface but lack the depth that makes the work actually work.
The doorman story keeps coming back for a reason. It asks a simple question:
What’s the part of your work that looks trivial from the outside, but is actually holding everything together?
No Comments