Picture the meeting. The design is nearly finished, polished, and laid in front of the client like a roast. The client studies it, nods gravely, and begins with a compliment, because the compliment is the anaesthetic. "Well done, guys, this looks great." A pause, to let the anaesthetic work. "But could we just change a few things?"

I was in that room. I started building digital experiences on CD-ROM, then on the web, and the method was the same on both: you began with a visual layout very close to the thing you were about to build. Not a sketch. The product's face, already made up, already smiling. Once it was signed off, every change request was a job in itself, always introduced by the same flattery. Sometimes the changes were so deep that what came out was a completely different design, still wearing the old one's folder name.

When a refined idea took days to make and days more to argue about, you learned to decide before you built. A cheap drawing to be wrong on, so the expensive one could be right. In my reading, a witness's and not a historian's, that is a large part of why the UX phase exists. A sensible answer to a price. Insurance against rework, sold as a philosophy.

Then AI arrived, and the price fell off a cliff.

Today a refined interface, close to final, takes hours. Not a week. Hours. Several versions, side by side, before lunch. When mistakes are nearly free, the insurance premium looks insolent.

Only I know what is happening on the other side of the screen.

A phase that lasts long enough stops feeling like an answer to a price and starts feeling like tides, or gravity, or the office coffee machine. So it will be defended, with the one argument that has held the gate for years: Only I know what is happening on the other side of the screen.

I admire that sentence. It needs no evidence, since nobody else can check. It needs no results, since the results are always the user's fault. Expect it in several editions: the solemn open letter, the petition, the street protest with a magic wand on every banner. I watched that sentence get promoted. More than once. The title never had a door. I knew account managers who became UX designers. I knew QA people who became UX designers. In one case I knew a receptionist who, between one quarter and the next, with no visible trace of method, became a sorceress of user experience, issuing verdicts on human behaviour with the authority of a taxi driver on the economy. (No offence to receptionists, usually the only people in the building who know what the customer wants. The point is that the title had no gate, so the authority had to be performed.)

And performed it was. For a depressing number of them, the real craft was talking about the user convincingly, as if they had a portal to his skull, or a crystal ball on the desk. The user feels abandoned by the carousel. Who could object? The user was not in the room. The sorcerer was.

The merely average designers found their own valid way to keep the job, too: make life hard for the ones with real talent. Overnight, the little boxes they had scattered across the layout became scripture, to be followed to the pixel, or else the user would be confused and stroll out of the funnel with one leg already inside. The gifted designer with a better idea was told it would confuse the user. See above, re: the room.

Then came the vocabulary. A new term every month, delivered like a discovery, photocopied by every other wizard in town within the week. And at some point, in meeting after meeting with the client, everything became a paradigm. The navigation was a paradigm. The checkout was a paradigm. Moving a button eleven pixels was, quite possibly, a paradigm shift. A word nobody can define is a word nobody can challenge. If the client frowned, you did not explain. You upgraded to "paradigm".

Newspaper-style cartoon: a sorceress pulls the word PARADIGM out of a top hat in a meeting room while a bored client rests his chin on his hand, a receptionist with a wand beams, and a gifted designer with a better idea is shushed by tiny wizards.

None of this means the user stopped mattering. That would be stupid, and also lazy, and I am only allowed one of those per article. Watching real people use the thing is still the best information anyone gets, and it just got easier: an afternoon, not a quarter. A machine can draw you forty screens. It cannot sit in the chair of the person who cannot find the button. What does not survive is two months of phase, upstream of everything, justified by the price of being wrong. Wrong has been marked down.

So the casualty is not a profession. It is a monopoly on a stage, and a toll on the way in. The people who cared will keep caring, with a sharper pencil and faster feedback. Those who mostly charged for the gate will notice, correctly, that the gate stopped being necessary.

They have had their stage. The rest of us have things to build.