01The Amplifier · part three

The thing that made it compelling was not the dogs.

It was that you could check it. Hundreds of episodes, books, years of audiences — and in every one the same shape. And the promise underneath was better still: this is available to you. There’s a method, it’s learnable, and it works.

02So what happened

We did not.

03What we did instead

We accepted the dog we have.

Not grudgingly — lovingly. It’s a good dog. It doesn’t do the things we hoped it would do, and it’s not as calm or as happy as we imagined when we brought it home. And underneath that acceptance is a quieter thought, which almost nobody says out loud: it’s probably too late now.

04

It failed because the work it implied was ours, and the story looked like it was about somebody else’s dog.

Every part of that is now true about AI, in a quarter of the time

05Interactive · the same ask, twice

One ask. Two approaches.

The owner on the left keeps looking for a better service. The one on the right wrote down three rules.

The ask: write the update for Thursday's leadership meeting.

Owner one — find the right one

Owner two — write down the rules

The services are real names and the outputs are illustrative. Nothing here is a benchmark — it’s the shape of what happens when the only thing you change is which box you type into.

06Why the left column does not move

The same competent nothing.

They’re working from the same absence. Not the right method — the right service. The one that will finally understand us out of the box. We try the new model, we read about the one with better memory, we move our work across and set it all up again.

07Everybody watched this and nobody took it away

The owner is the one getting the instruction.

10mHow long the dog takes to change.

∞How long the owner takes. Stand like this. Say it once. Mean it. Don’t reward the thing you’re complaining about.

So the reason the country isn’t full of well-trained dogs isn’t that the method was hard, or fake, or unavailable. It’s that the method required the owner to change something about themselves, on purpose, and keep it up.

08

He was not teaching dogs to obey. He was teaching people to be consistent enough to be understood.

And that sentence, with two words changed, is the entire product

09Interactive · the objection

“But it’s too late for mine”

Drag it. The answer doesn’t move, and that’s not a bug in the toy.

Time spent re-explaining yourselfJust started

Today.

Best possible moment. Nothing to undo and nothing to admit.

Seven years of doing it the hard way isn’t a reason to keep doing it the hard way. It’s seven years of evidence that the hard way doesn’t improve on its own.

10Nearly one to one

It’s the training — and it’s training for the person.

The dog→The assistant. A good one. It will not become a different animal because you found a newer one.

Looking for the right dog→Auditioning services, hoping one turns out to already know you. It was never coming.

“Stand like this. Say it once.”→The small things the owner has to do the same way every time.

The owner changing→The part everybody watched and nobody took away. The dog changes in ten minutes; the owner is the one getting the instruction.

Writing it down→An interview about how you actually work, once. A plain-text file you own — so the consistency does not depend on you remembering.

Stop looking for the service that already knows you. Start knowing yourself accurately enough to tell any of them.

11The same week either way

One of them ends with a file.

The people getting real leverage out of AI aren’t a different kind of person. They’re not more technical and they didn’t find the right provider. They wrote down what they know, and now every tool they pick up starts from it.

Take the mirror test How Workbrain works