Almost everything important about a business cannot be measured. The London hotel that fired its doorman is about to learn this lesson again, on a much larger scale, because of AI.
Rory Sutherland, the British advertising thinker, often tells of a London hotel that decided to cut costs by letting go of its doorman. The math was straightforward. A doorman is a salary line. His job description was simple: open the door. Doors can open themselves. So by Friday he was gone, and the spreadsheet looked cleaner.
Then the complaints began. Guests were getting wet in the rain while waiting for taxis they could not hail. Returning customers were not being greeted by name. The lobby felt colder, in the way a room feels colder when no one is paying attention to whether you are comfortable in it. Within weeks, the hotel understood what it had done. The doorman’s job description was “open the door.” His actual work was something else. He was the first human being a tired traveller encountered after a long flight. He hailed taxis with an authority the average guest could not summon on a wet London evening. He remembered names. He noticed when a regular looked unwell. He held the social temperature of the entrance.
None of this was in his contract. None of it was on the spreadsheet. And so, when the spreadsheet came for him, none of it was defended.
Sutherland calls this the doorman fallacy. I think it is the single most expensive mistake modern businesses are making, and it is about to get worse.
The fallacy is what happens when you define a job by its most visible task and dismiss everything that surrounds the task as ornament. The visible task is what shows up in the job description. The work that surrounds it is what makes the visible task matter. They are not the same thing. They have never been the same thing. But the spreadsheet, by its nature, can only see the first. And when a business is run primarily through spreadsheets and dashboards, it begins to systematically destroy the second.
Here is why this matters now. Artificial intelligence is about to make the doorman fallacy worse than it has ever been.
Almost every business is currently sitting across from a consultant, or a board member, or an internal team being told that AI can do the visible task of every job in the company. The consultant is somewhat right. AI can write the email, run the ticket triage, draft the quarterly business review, produce the renewal forecast, generate the sales deck, summarise the call, build the dashboard. The list grows every quarter.
This is the part everyone is paying attention to, and they should. The visible task of most knowledge work is being automated in front of our eyes, and any business that does not put AI to work on those tasks will lose to the ones that do. The doing is becoming cheap. That is not a bad thing. It is a permanent change in the economics of how work gets done, and the only sensible response is to embrace it.
But the lesson of the London hotel is that the visible task was never the whole job. And it is about to become a smaller fraction of what a human being is paid to bring to a business.
What AI cannot do is the work that surrounds the visible task. The work of noticing. The work of holding the social temperature. The work of being a body in the room when a customer needs to feel that someone has heard them, not processed them. The work of carrying institutional memory in a way that lives in a person, not a Notion page. The work of reading a quiet email from a long-term client and knowing, from the cadence alone, that something has shifted. None of this shows up in a job description. None of it can be defended in the language of dashboards. And yet, in the AI era, this is the work. It is no longer the part of the job that surrounds the value. It has become the value itself.
The smartest companies in the world are already operating on this assumption, though most have not put it in these words. They are automating the visible task ruthlessly. They are also, simultaneously, investing in the part of their workforce that does what the spreadsheet cannot see. Both moves at once. The two are not in tension. They are the same strategy. You free your people from the doing so they can spend their time on the being. In the end, knowledge work is not just production: it is presence, judgment, and attention.
This is also where the doorman fallacy is going to claim its biggest victims over the next three years. There will be leaders who look at AI’s ability to handle the visible task and conclude that the role itself is now redundant. They will fire the customer success manager because the AI can run the QBR. They will close the support team because the chatbot can triage tickets. They will lay off the community manager because the AI can post on the channels. For a quarter, maybe two, the spreadsheet will look cleaner. Then the customers will begin to drift. The product team will start flying blind. The senior people will find themselves doing the work the junior people used to do, and doing it badly, and resenting it. By the time the data shows the loss, the doorman will be long gone, and no one will remember he was the one keeping the lobby warm.
At Mozisha, we believe that technical fluency is becoming table stakes, and our operators are AI-fluent by design. They use the machines to clear the doing off their desks, every day, so nothing visible in their job takes more time than it has to. That is the floor.
What we work hardest to develop is the ceiling. How to be present with a customer who has gone silent. How to hold a difficult conversation without softening it into mush. How to write an email that registers, in its tone alone, that you understood not just the question but the worry behind it. How to read the temperature of a Slack channel and know when a small intervention will prevent a large rupture. These are not soft skills. They are the highest-leverage skills a knowledge worker can bring to a business in the AI era, because they live in the human relationship, not in the task. AI can imitate the task. It cannot stand in the relationship.
When a global company hires through us, they get two things at once. An operator who has built genuine fluency in the tools, so the visible task of their role gets done faster and cleaner than it would otherwise. And an operator trained to spend the time AI gives back on the part of the work the spreadsheet cannot see. The doing and the noticing, in the same person. The execution and the attention, at the same desk.
This is the value of a human being in a world where the doing has become cheap. Not less human because AI is doing more. More human, precisely because AI is doing more. The doing is the floor. The noticing is the ceiling. The businesses that survive this decade will be the ones whose teams can hold both at once.
The lobby is colder than it used to be. Most leaders will not notice for a quarter or two. By then, the customers will have started to drift. The dashboards will show churn. The spreadsheet will show efficiency.
No one will remember who kept the lobby warm.
It was never the door. It was the doorman.
