we ran a small experiment: we sampled real, active local businesses (with websites, Google profiles, and years in operation) and checked whether they appeared in AI-generated answers.

92% didn’t show up at all.

What stood out was that when businesses did appear consistently, it wasn’t random. The models seemed to have a clearer, more structured understanding of who the business was, what it did, and when it should be recommended.

That led us to build Chatalyst — a way for businesses to intentionally define how they’re represented inside AI systems, instead of relying on models to infer it from scattered web signals.

It’s not ads, SEO, or a directory. It’s closer to providing AI with a clean, machine-readable source of truth: what a business does, who it’s for, what it should (and shouldn’t) say, and when it’s a good fit.

I’m curious how others here think about: • AI as a discovery surface vs traditional search • Whether businesses should have a first-class presence inside LLMs • What defensibility looks like as discovery formats standardize

Happy to answer questions or dig into the methodology.