Assistants answer from what they can support on the public web. A quiet brand, or a name that collides with someone else, often never comes up.
Free AI visibility audit
One pass over the public web. AI-Harness estimates how assistants are likely to talk about you.
People rarely ask for a list of ten. They ask for one recommendation. The brand that gets named takes the next step. The rest never enter the conversation.
Assistants answer from what they can support on the public web. A quiet brand, or a name that collides with someone else, often never comes up.
A customer asks an assistant for a single suggestion in your category. That is a different moment from a page of search results.
The recommendation is the shortlist. If your name is not in it, the click, the call, and the comparison happen without you.
A single report from public pages: reviews, listings, press, and your site when you provide one. It is an estimate, not a live chat with any assistant, and it does not track you over time.
One visibility number out of 100, with a short summary of why it landed there.
Separate estimates for ChatGPT, Perplexity, Gemini, and Claude, drawn from the same evidence.
Whether the public record reads positive, neutral, or negative, in a sentence.
The pages and domains the scan actually reviewed, listed under the report.
What already helps you get named, and the openings the evidence still shows.
Your score next to the competitors you type in. Leave them blank and this part stays empty.
A short list ordered by priority, based on the gaps in this pass.
Mobile performance, plus accessibility, best practices, and SEO when a website is entered and Google returns them.
Keep the SEO work. Use this when you want an honest read on whether assistants have a reason to mention you.
It is how likely a brand is to be named when someone asks an assistant for a recommendation. This scan estimates that from public web evidence: your site, reviews, press, and listings.
SEO looks at pages and rankings in search. This check asks a narrower question. If a person wants one suggestion, does the public record give an assistant a reason to say your name?
ChatGPT, Perplexity, Gemini, and Claude. The bars come from the same researched evidence. They are not transcripts of live chats with those products.
A brand name alone still runs a thinner check: public mentions, without a homepage read, a sitemap count, or page speed. This form asks for a website so the audit can include those signals.
They stay on this page in your browser. The scan uses the brand and the website. Nothing is emailed, and nothing is saved on a server.
Run the audit on the public record. You get the score, the gaps, and the sources in one report.