AI knew the company. It still didn’t recommend it.

I ran into a result while testing Riseklix that changed how I think about “AI visibility.”

We tested a real company across several buyer situations.

When the company was explicitly named, the AI systems recommended it in 9 of 12 checks.

Then we removed the company name and asked about the same underlying buying needs.

0 of 24.

The strange part is that the models clearly knew the business.

They understood what it sold.
They understood its capabilities.
They could talk about it when prompted.

They just didn’t retrieve it naturally when someone needed what it sold.

I think those are two very different forms of visibility.

Search gave us a fairly intuitive model:

You rank #3. Someone else ranks #1.

AI discovery is messier.

A buyer can describe their company, problem and constraints, ask who they should consider, and receive three names.

There may be no second page.

So I’m starting to think the interesting question isn’t:

“Does the model know my company?”

It’s:

“What situations cause the model to think of my company without being reminded?”

That’s why I ended up organizing Riseklix around buyer situations rather than a giant fixed list of prompts.

Still early, but this distinction has become one of the more interesting things I’ve seen while building it.

Curious how other makers think about this:

Is recognition enough to count as AI visibility, or should visibility only really count when the company gets retrieved unaided?