"The first principle is that you must not fool yourself, and you are the easiest person to fool."
Richard Feynman

We sell what we know. An operating system for OKRs: 124 linked concepts, ten years of practice, licensed to companies who want goals that move things. Which makes us exactly the kind of business AI came for first.

So this year we ran the experiment most knowledge businesses are quietly avoiding. We handed a stranger our floor plan and timed the break-in.

The stranger was a fresh AI. No context, no access, no memory of us. The floor plan was everything a visitor to our website sees: the page copy, the fourteen notes we publish in full, and the concept map, all fifty-four public titles and how they connect. We publish that skeleton deliberately. It is our marketing.

Its job was to rebuild the licensed methodology from that alone. An independent grader scored the reconstruction against the real thing, note by note, then checked for contamination. Clean.

The result: 68 out of 100, in about ten minutes.

Now the part I would rather skip. Before the experiment I predicted 80. Wrong in the strange direction: I overestimated the machine while actively trying not to. That is why you run the experiment. The vibes lie both ways.

And the split matters more than the number. Our beliefs cloned at around 90. Read the copy's philosophy and you might hire us. Our machinery graded 45: wherever the method specifies a real instrument, the burglar left a substitute from the industry average. We grade quarters on judgment, where 50 percent progress can be excellent if the learning was worth it. The copy scored 0.0 to 1.0 and called 0.7 success, the exact convention our method exists to kill. The grader needed one line: "drifting toward generic Google-style OKRs precisely at the points where this methodology differentiates itself."

We stopped guarding the floor plan and started counting what the burglar could not carry. The free and licensed boundary is now drawn on purpose. The install became a product. And every engagement now feeds evidence back into the system, because the one thing no break-in can take is what only happened to us.

Two lines for the notebook. Your marketing is their training data. Your table of contents is training data too.

If a competitor handed your public footprint to a model this afternoon, which parts of your business would clone at 90, like our beliefs, and which would grade 45, like our machinery? If you do not know, that is the answer.

Run the experiment.

See you in a fortnight,
Matt

P.S. The concept map the machine copied is open at the OKR Knowledge System. The parts that cloned are free to read. The parts that did not are what we license. And the free AI OKR Coach kit lives here too, no charge, nobody will call you.