AI Accuracy
Every factual claim the engines make about you, checked against your live Google listing.
You have never read a word of it. Truesay shows you every claim ChatGPT, Claude, Gemini and Perplexity make about you, proves which ones are wrong, and fixes the sources they read.
We asked the engines who they recommend across five industries, then ran the same searches on Google. The businesses AI recommends are mostly not the ones ranking.
Three to five businesses per answer, built from whatever the web says about you. Often wrong. Often without you.
“Best boutique hotel in Austin for a weekend?” Millions of high-intent buying questions go to AI every day.
It recommends a handful of businesses in a sentence. No ten blue links, no second page. You’re named, or you’re invisible.
Wrong price, stale hours, or simply absent, and the customer books your competitor without ever seeing you.
Ranking gets you into the pool an assistant draws from. It does not decide who it names.
In dental and hotels, closer to nine in ten of the businesses AI names were nowhere on page one and only one of them is being measured. See how we measured it →
We measure, we write the fixes, and we confirm the ones that stuck. Your part is approving them and pasting a few lines, which takes minutes a month.
We ask ChatGPT, Perplexity, Gemini and Claude the real questions your buyers ask in your market, then record exactly what they say about you.
You get your Answer Share and a per-engine breakdown of where you win, where you’re only mentioned, and where you go missing entirely.
We write the corrections for the facts AI reads: your Google Business Profile, reviews, the sources that get cited, plus a Facts Page we host outright. You paste what needs pasting, we handle the rest.
We re-read the live source to confirm each fix is genuinely serving, then re-test. A gain only reaches your timeline once it clears the noise band twice.
We don't hand you a tool and a to-do list. We work every surface that decides what AI says about you: measure it, write the fix, then confirm the fix is live.
One number for how often AI names you, per engine, with the noise band it is measured against.
Every factual claim the engines make about you, checked against your live Google listing.
A verified, schema-marked record engines can cite, plus your Google Business Profile put right.
We correct the third-party pages that get you wrong.
Your own site, read with JavaScript switched off, the way the engines read it. Plus whether you are accidentally blocking them.
Every fix re-tested, and only logged once the gain survives a second measurement.
Your facts in a form a machine can call, so an AI agent gets your hours and your price from you rather than guessing from a directory. Tripadvisor and Eventbrite already have one. Almost no independent business does. On every paid plan.
This is the product behind your account. You don't have to run it, but you can watch every move.
One number for how often AI names you, per engine, always shown with the date it was measured and the noise band around it. No guessing, no vanity metrics.

We ask the 40 questions your buyers actually ask, and record which ones name you. The families you are absent from are where the work goes first.

Every fix is re-tested, and a gain is only logged once it clears the noise band and survives the next measurement. Proof, not promises.

As AI agents start reserving and buying on people’s behalf, Truesay gives you a machine-callable endpoint so they can find your facts and hand off a booking to your own site.

One reads it in a browser. The other fetches it once, with no JavaScript, and leaves. We read your site the second way and score two things: can an assistant parse your prices, hours and policies at all, and can an agent then act on them. A site can pass one and fail the other. Fail the first and you are invisible. Fail the second and the agent hands your booking to whoever it can transact with instead.

Some tools pump 30 AI blog posts a month onto your site and hope. That is what AI is learning to ignore. Here is what GEO actually involves.
We have no customers yet, so there is nothing to quote. What we have is the measurement, published with its method and its date, including the parts that do not flatter us.
| Engine | Claims checked | Verifiably wrong | Error rate |
|---|---|---|---|
| ChatGPT | 35 | 0 | 0% |
| Gemini | 5 | 0 | 0% |
| Claude | 6 | 1 | 17% |
| Perplexity | 36 | 11 | 31% |
Measured 20 September 2026. What counts as an error was fixed in writing before the data was seen: a rating out by 0.2 stars or more, a review count out by more than 15%, a phone number that differs, opening hours that differ on a named day. One category, one city, and the engines answered very different numbers of checkable questions: 36 for Perplexity, 35 for ChatGPT, 6 for Claude, 5 for Gemini. A rate over five questions is not the evidence a rate over thirty-six is. How we measure.
No. Truesay fixes the surfaces engines read, your Google Business Profile, a hosted Facts Page, the sources that cite you. You don’t publish anything or touch your site.
SEO ranks links; content tools flood the web with articles. Truesay changes what AI assistants actually say when they answer a buyer in a sentence, by correcting your facts rather than adding noise.
We measured this rather than guessing. Across every market we tested, 77% of the businesses AI recommends did not rank on Google’s first page with their own website. In dental and hotels it was closer to nine in ten. Ranking still helps, because engines read the web, but the two lists have drifted apart far enough that one no longer tells you about the other. Your free report shows you which side of that gap you are on.
Yes. GEO (generative engine optimization) and AEO (answer engine optimization) are the industry names for getting named inside AI answers instead of ranking a link. Truesay is that, done for you and measured rather than guessed, we run the buyer questions, report who gets named, fix the facts behind it and re-measure to confirm the change was real. If you want the longer explanation, we wrote one.
ChatGPT, Perplexity, Gemini and Claude, the assistants your customers already use.
The work, verified. Each month we prepare four fixes, and once you apply them we check by machine that each is genuinely live. If we do not deliver four, or one we marked live is not, that month is free. We do not guarantee an Answer Share number yet, because we have not published a controlled result proving our fixes move it. We would rather promise the part we control.
Never. We only publish your true, sourced facts and nothing goes live until you approve it.
Run your free report in about a minute. No account, no setup, just the truth about how AI describes you today.