About forty salaries, once. 18 agents, today. Same founder. I did not gain speed. I deleted four steps out of five. That is the whole difference. Most companies buy AI to run the same steps faster. Five steps, faster, is still five steps.
I did not plug AI into my old company. I started from the blank page.
Step to One came first. You get a photo, straight from the software. One click. The average consultant teaches teams five steps. Then the same five, faster, with AI. Step to One deletes four of them. Nobody does them faster. Nobody does them at all.
Then the workshop, built with agents. Some agents build. Others verify. I expected the verifiers to rubber stamp. Nope. Every verifier found a real defect. Not a style remark. Today the suite shows 248 green tests.
The honest ledger now. Step to One revenue: not yet. Workshop revenue: not yet. Payroll saved, in euros: not yet. Headcount: 18 agents, zero salaries.
One line from the other column. A scanner bought at 60 000 €. It billed 17 000 €. Capacity bought before anyone could absorb it. A licence nobody opens is the same purchase.
Sears put its catalog on the internet. Same products. Same house. Same organisation. The medium changed. The company did not.
Every new tool first copies the old shape. The agent replaces the clerk. The org chart stays. Then someone starts from the blank page. That someone competes on absence, not speed. Fewer steps. Fewer handoffs. Fewer places to fail.
Everyone can rent the same shovel. Everyone can rent the same model. The shape of a company is not for rent. It carries titles, habits and budgets.
If you bought AI last quarter, ask one question. Which step disappeared? If none, you bought speed. Speed on a step that should not exist is waste. Faster waste, but waste.
Five steps, faster, is still Sears. My photo arrives in one click.
First published on X.