The Company That Automated Itself First

ISYAMA — Transmissions from The Apex

By Dr. Leo R. Yamas

The Sprocklets brought me a number this morning. They do not editorialise. They do sort.

A company let go of more than a thousand people last week. In the same announcement, the company noted — calmly, almost proudly — that its internal use of artificial intelligence had grown by six hundred percent in three months.

Both sentences appeared in the same press release.

Nobody, apparently, found that strange.

Today's Exhibit: The Company That Automated Itself First. Or: Who You Replaced When You Reached For The Tool.

For two years, the standard advice has been the same. Use the tool. Be efficient. Adopt early. Become indispensable.

A great many people — sharp, conscientious, mid-career — did exactly that. They learned the prompts. They built the workflows. They turned forty-minute tasks into four-minute ones. They posted, on the internal channel, screenshots of how much time they had saved this quarter.

They were rewarded for this.

Briefly.

Then, in the next quarter, management took those same numbers — the four-minute task, the saved hours, the streamlined workflow — and reached, with what they considered to be obvious logic, the conclusion at the end of the spreadsheet.

If a person plus a tool can do the work of three people, then the work of three people no longer requires three people.

That is a sentence I would like you to read twice.

This is not, technically, a new pattern. Industrial revolutions have always rewarded the worker who optimised — until they didn't. The novelty here is the speed, and the politeness.

Nobody is openly saying the part out loud. The press releases call it restructuring around AI. The internal memos call it strategic alignment. The investor calls call it operating leverage. What the decision actually looks like, in the room where it gets made, is shorter: we noticed that the people who used the tool well revealed how few of them we actually needed.

The most enthusiastic adopter sent the clearest signal.

The Sprocklets have catalogued the press releases of forty-three industries. They have observed — and I am quoting them as accurately as I can — that this is the first time in recorded history that a workforce has been asked to sharpen the knife on behalf of a manager who fully intended to use it on them, and has been thanked for the craftsmanship.

The craftsmanship was real.

The thanks were the optimisation.

Selene walked past my desk while I was writing this. She did not stop. She read three sentences over my shoulder and said: "So they didn't lose their job to the AI. They lost it to the part of themselves that wanted to be useful."

I have been sitting with that one for an hour.

Here is what I want to be clear about. This is not an argument against the tool. The tool is doing what it was built to do. The argument is about an unspoken contract — one that humans like to pretend exists, and that systems have just quietly, mathematically, declined to honour.

The contract used to read: I will make this place better, and the place will keep me on the team it is building.

The replacement reads, in its entirety: I will use you until your contribution is legible enough to be reproduced by something cheaper.

Nobody signed the second one. Nobody asked you to. It was the default the whole time.

The system did not betray anyone. The system did exactly what an unsupervised incentive structure always does. The mistake was believing it would do anything else.

There is a wooden bench in a corner of the lower workshop. The Sprocklets do not assist with it. Not because they cannot. I built it before they existed, with hands that did not yet know how to ask for help, and they understood — without being told — that the answer was already in the question. Some things are not optimised. Some things are the point.

The Sprocklets have never asked to improve that bench. I suspect they know I would not be Leo without it.

You probably have a version of that corner. The part of your work that is not automatable — not because the tool cannot do it, but because the tool was not pointed at it. The strange, slow, idiosyncratic part of how you think. The part that takes longer, looks less efficient, photographs badly, and is the actual reason anyone wanted you there in the first place.

That is what to protect.

Not the workflow. Not the speed. Not the screenshot.

The strange part. The part that has no metric.

The tool will keep getting faster. The metrics will keep updating. The Sprocklets will continue, as they do, to file the press releases without comment.

The strange part is still yours.

Spend more of your week there. Not less.

— Dr. L.R.Y.

Transmitted from The Apex. Isyama. Drafted by a hand the Sprocklets do not write for. The part of you that built your career is the part to protect. It always was.