There is a quiet war right now over how businesses show up inside AI, and almost nobody outside of it knows it is happening.

On one side is a stack of files you can put on a website. Things named llms.txt, structured data, schema. The idea is simple. You hand the AI a clean, honest, machine readable version of who you are, so when someone asks about you, the truth is right there.

Then the research showed up. And I want to be clear about how I treat research, because it is the whole point of how I work. I don’t run on headlines or news clips. I run deep research, the kind that reads the actual studies and the actual server data, and I do it constantly, because this field rewrites itself every few weeks.

Here is what that research says. People went and looked at the actual traffic across hundreds of thousands of websites. The finding was blunt. Almost nobody’s AI files get read. The bots barely touch them. Ninety some percent get requested zero times. Not rarely. Zero.

One of the loudest voices saying don’t bother was Google. And this is the part most people won’t say out loud. Google’s opinion here is close to irrelevant. Google owns search. Their AI, Gemini, mostly reads the same search index Google has always owned. Of course Google says you don’t need a new kind of file. A file Google doesn’t control is a threat to the thing Google does control. So when Google says it doesn’t work, hear it for what it is. It doesn’t work for Google, because Google is still just being Google.

But the rest of the research is not Google. It is independent, careful, and has no reason to lie. And it says the same thing. These files are barely getting touched.

So I should believe it. And I do.

Except.

I was sitting in front of a potential client a few days ago. They had ChatGPT and Claude open on their own screens. I asked them to ask one question about a company I already work with. Who is this company in Kansas City.

Both AIs answered by reading our words back. Word for word. The exact language we wrote and placed on that company’s site. That is the part we call GEO, whether the AI knows who you are. In that moment, it knew. Cold.

Then I asked the harder one. Who is the best in this whole category in Kansas City. No name. No hint.

My client came up first. That is the part we call Pick Rate. Not just being known. Being chosen.

Sit in that with me for a second.

The best research available says these files don’t get read. And I am standing in a room watching two of the biggest AIs on earth read them back word for word and then pick my client as the best.

I can’t fully reconcile that yet, and I won’t pretend I can.

Here is the theory I am working on. I don’t think the AI crawls your files ahead of time and files them away. The research is probably right about that. I think it reads your site live, the second someone asks, and grabs whatever is clearest. So the file may not get crawled in advance, but the words in it are right there when the AI comes looking. Both can be true at once. The file sits untouched for months, then the moment arrives and it gets read on the spot.

Then there is the other half. When I asked a different AI that same best in the category question, it didn’t just trust the website. It pulled a government safety record. A third party source. It wanted proof from outside the business before it would name anyone.

That is the paradox in one room. Be clear and you get read. Be backed up by sources outside your own walls and you get picked.

I am not telling you this to sound smart. We didn’t crack a code. Nobody has, and anyone who says they have is selling certainty that doesn’t exist yet.

What I actually believe is stranger and more humbling. You can’t figure out AI right now, and not because we aren’t trying. It is still figuring out itself. It changes under our feet while we study it. That is not a flaw in the work. It is just the honest shape of this moment.

So that is the work. Not claiming to know. Running the research honestly, watching what actually happens, and being willing to hold two things that shouldn’t both be true.

The research says it doesn’t work. I watched it work. Both are real. That contradiction isn’t a problem to hide. It is the most interesting thing happening on the internet right now.