We also want to clear up a naming confusion early, because it matters for anyone trying to find us again after a first conversation: our product used to be discussed internally under a different working name, and in English-language contexts that name reads as something else entirely. In our own testing, that older name got answered by AI models as a GPS fleet-tracking app and as a Gmail email-tracking browser extension — neither of which is us. So in English, the product is MaxGrowth Answers, full stop. The company behind it is 北京口袋智创科技有限公司, which we keep in its original Chinese legal form in English text because it has not been independently verified in a romanized form, and we'd rather be precise than convenient.
Real questions, live engines, and who gets named instead
The audit is not a keyword-rank checker repurposed for AI. It runs a set of real questions — the kind a prospective customer would actually type into an AI assistant — against the assistants themselves, and records what comes back: whether your brand is mentioned, whether it's recommended, what sources the AI cites, and who gets named instead of you. That last part is usually the most useful, because it's rarely a direct competitor. In our own testing across ChatGPT, Gemini, and Perplexity, the models frequently answered with the wrong company altogether — different similarly-named businesses depending on which engine you asked, including one confusion that resolved to a company in an entirely different country and industry. Each engine got confused in a different way, which is itself a finding: there is no single "AI ranking" to fix, there are several independent surfaces that each need their own read.
Live runs, a stated layer, and denominators that hold
We built the audit around actually running the questions live rather than inferring visibility from search-engine proxies, because the two don't correlate the way people assume. As one concrete data point from our own measurement work: we ran 32 real questions against ChatGPT, Gemini, and Perplexity — 96 test runs in total, zero failed to return an answer — and repeated the same 32 ChatGPT questions again a few days later as a consistency check, again with zero failures. That's the API-response layer, not what a human sees scrolling a chat app on their phone, and we say so explicitly in our own reporting because the two surfaces can differ.
The same discipline applies domestically: in a separate batch covering 53 questions across three Chinese AI platforms, we logged 159 individual responses, and even the one response that failed to return cleanly was kept in the denominator rather than dropped, so the count doesn't get quietly inflated by excluding inconvenient results. Every platform we test cites sources differently — one surfaces search-style reference links, another embeds citations directly into the answer text, a third assembles its own retrieval before answering — and we do not add those numbers together into a single "total mentions" figure, because they are not measuring the same thing. If a report ever hands you one combined number across engines, ask what happened to the differences between them.
A diagnostic read, including the zeros
The free audit gives you the raw read: which of your real customer questions get answered with your brand present, which get answered with someone else's, and which sources the AI is actually pulling from when it forms that answer. It is diagnostic, not promotional — we are not going to tell you a number improved unless we can show you which specific question and which specific engine that number came from. If your brand doesn't show up at all in the results, that's also a legitimate finding, and a common one; AI visibility today is still mostly wide open, not owned by anyone.
A standalone diagnostic, and what we won't promise
AI visibility work sits alongside our other three lines — organic community engagement overseas, social comment-section reputation upkeep, and this kind of AI visibility monitoring — but the audit itself stands alone as a diagnostic. You don't need to be running any of the other work to get a useful read from it. What we'd ask in return is patience with precision: we're not going to promise a ranking outcome or a timeline for one, because AI answer generation isn't something any outside party controls end to end, us included. What we can promise is that the numbers we hand back came from questions we actually ran, on engines we actually queried, with the failures counted rather than hidden.
If you've talked to us before under the old product name and are trying to find this page again: same team, same underlying work, corrected name. Everything above is what "free audit" concretely means here — not a sales teaser, the actual measurement.
Want to see how AI engines describe your brand today?
Get a free growth audit