Somebody on this island gets up before the sun and asks the machine a question. “The Best Hotel in New York City.” One run. No retries. Whatever it says, it says. Then he does it again the next morning. And the next. It is not glamorous. Nobody watches. It is just the work.
Everybody else is doing something else entirely.
Ask any AI that same question and it names a few hotels. Every luxury hotel in New York City wants to be one of those names. So a crowd of companies showed up selling GEO tools. Generative Engine Optimization. Big words, simple product. They check whether the AI mentions you. Visible or invisible. Green light or red light.
Here is the crowd. Lighthouse sells software that tracks how you rank in AI answers, and in May it bought a company called Hotelrank.ai to track it better. Operto has a free tool that tells you if you appear. HotelWorld AI publishes a study on who is visible and who is not. Curacity watches too. LSEO, O’Rourke, Lotus Marketing, Hotelwize, Eighty Days, and hotelmarketer.pro sell the same promise in different wrapping. Make your site “AI-ready.” Add some tags. Cross your fingers.
Read the list again. Almost every GEO tool on it does the same single thing. They tell you whether the machine sees you.
A GEO tool is a mirror. A mirror tells you there is a stain on your shirt. It does not clean the shirt.
Now, a few of these GEO tools will say they do more than watch. They say they help the machine understand you. They tidy up your data so the machine can read it. They hand the machine a clean business card. One will sell you premium placement inside a booking app that lives in ChatGPT. That is shelf space. Shelf space is fine. Shelf space is not the answer.
Every pitch on that list ends the same way. Do this and the AI will recommend you more often. That is the claim. It is the only reason anybody writes any of them a check.
So we went looking for the verification of that claim. Not the promise, the verification. A property named. A date. A before and an after. One screenshot a buyer could find and check without having to ask anyone’s permission.
For every company named above, what we found published was that company’s own announcement about itself. We may have missed something, and we would rather be corrected than be wrong. So do not take our word for it. Search any name on that list and see what comes back. It takes ten minutes.
Now the harder question, and it is the only one that matters.
When the machine decides which hotel in New York is best, where does that answer come from?
Not from your business card. A machine does not form an opinion out of your room descriptions and your amenity list. It repeats a judgment that somebody else already made and published. Somebody has to be the source of the ranking. Somebody has to have said it first.
Nobody in that list up there is building that source.
We live alone out here. The competition would visit. They can’t find the boat.
Because that source is the only thing we build. We published The Best Hotels in New York City at seven minutes past midnight on July 10, 2026. It had no masthead, no history, and no century of print behind it.
Three days later it was sitting in second place on Google, above Condé Nast Traveler and Forbes Travel Guide.
Two days after that, Google stopped listing links and answered the question itself. It said which property ranks among the best in the city, and for that judgment it cited us, above Forbes. It quoted our dated ranking back as its source material. The machine gave our answer.
We published all of it, including the mornings it slipped. Every screenshot is dated and anyone can check it: A Three-Day-Old Page Outranked Forbes and Condé Nast
So picture the field. A crowd on the shore with binoculars, shouting out whether they can see you. A few of them polishing your business card. Then one company, alone, writing what the machine says about you.
Nobody else is on this island.
Sit with why that is.
