Why we exist
Something changed in how people find a business, and almost nobody told the people it happened to.
A billion people a month now use Google's AI Mode — that number is Google's own, announced by CEO Sundar Pichai at Google I/O. Hundreds of millions more ask ChatGPT. When someone asks one of these systems who to call for a plumber, where to eat, which shop to trust, the answer comes back with one or two names in it. Not ten blue links. One or two names.
The businesses that don't appear in that answer usually aren't worse. They're the taquería with thirty years of regulars, the barber whose chair your father sat in, the plumber everyone in the neighborhood calls. They're invisible to the machine for a boring, fixable reason: nobody ever installed the plumbing that lets a machine read them — the structured data, the verified profile, the prices written where software can find them. Invisible not by choice, but by configuration.
Configuration is the whole point. It means this isn't a talent gap or a quality gap. It's an information gap — and information gaps are the unfair kind, because the only thing separating the businesses that show up from the ones that don't is whether somebody told them.
Large companies have agencies watching this shift daily. They were told in the first month. The businesses on the other end — family-run, one location, owner does the books on Sunday — are the last to hear the ground moved, and the most likely to get sold something dishonest when they finally go looking. This field is full of guaranteed-#1 promises and statistics nobody can trace to a source. We went and checked those numbers ourselves, and most of them fall apart.
That's the gap we exist in. Not because AI visibility is exotic work — it isn't — but because the people who most need it are the least likely to be handed an honest version of it.
Anyone can write a values page. The only test worth applying is whether a company does things that would make no sense if the stated belief were false. Here's ours, with the receipts:
| What we believe | What we did |
|---|---|
| You should be able to check our work | We ran our own scanner on our own website and published the failing grade — 69/100, then 71 — including a note explaining why the +2 wasn't real progress |
| A tool that grades others has to survive being graded | When our scanner briefly reported a false perfect score, we found it, published the mistake, and rebuilt the check |
| You shouldn't have to pay to find out whether you have a problem | We scanned 214 businesses across six towns — Napa, American Canyon, St. Helena, Calistoga, Fairfield, Vacaville — and gave every result away |
| The Spanish-speaking half of a town isn't an afterthought | 44 of our 59 pages — every city report and nearly every guide — have a real Spanish version written against Spanish search, not a translate button. The remaining 14 are on the list |
| Nobody should promise what they don't control | We have never guaranteed a ranking or an AI mention, because no one can. Our paid guarantee covers implementation — every fix done and verified, or it's free |
We won't tell you we can get you to the top of ChatGPT. Nobody can, and anyone selling that is selling weather. We won't quote you a percentage lift we can't trace to real research — the honest state of this field is that independent revenue data doesn't exist yet, and we said so in public even though it would be easier to borrow a competitor's number. We won't count anything you tell us about yourself toward your score; only what a machine can independently verify counts. And we won't pretend the scan is a favor that obligates you to anything.
The honest limit on all of this: we're a small company doing this in one valley, and we don't have a decade of outcome data to wave at you. What we have is a tool that shows its work, research nobody else bothered to do, and a habit of publishing our own mistakes. Judge it on that.
The plumbing that makes a business readable to AI is not expensive and not complicated. It's mostly free, mostly a weekend, and mostly nobody's job — which is exactly why it doesn't get done. Our whole business is closing the distance between "this is fixable" and "someone actually fixed it," and doing it in a way you can verify at every step.
Start by finding out where you stand. It costs nothing and takes about a minute.
See what AI can actually read about your business →
The AO Audit is a Napa, California company that measures how visible a business is to AI engines like ChatGPT, Gemini, and Perplexity, and shows exactly what to fix. It was founded by Alejandro Ojeda, whose name and face are on it, and who answers the email himself. AO stands for Answer Optimization — the practice the wider industry also calls AEO or GEO.
Because a billion people a month now ask an AI who to choose, the AI answers with one or two names, and the local businesses that earned their reputations over decades are the least likely to be in that answer — not because they are worse, but because nobody installed the machine-readable plumbing that AI systems depend on. Big brands have agencies for this. Small local businesses mostly have nobody, and are the most likely to be sold something dishonest when they go looking.
Because a business owner shouldn't have to pay to find out whether there's a problem. The scan runs on the real website, produces a real score with every check shown, and the full report costs nothing. Paid services exist for people who want the fixes implemented for them, but the diagnosis is never the thing being sold.
The fair way to judge that is by what a company does that would be irrational if it were lying. This one published its own failing grade twice, disclosed a bug where its own scanner briefly reported a false perfect score, scanned 214 businesses across six towns and gave the results away, publishes real Spanish versions of 44 of its 59 pages, and refuses to guarantee rankings or AI mentions because nobody can control those. Every one of those is checkable on this site.