Ask one question the way a customer would, and see whether ChatGPT, Gemini, Claude and Perplexity name your brand in the answer, what they cite instead, and what to do about it. Two engines free, no account.
What This Checks
You type a question, the name to look for, and optionally your domain. The check puts that question to the AI engines with web browsing switched on, the way a customer using them would get it, and reads each answer for your brand: named in the text, or one of your pages cited as a source. It keeps the full answer so you can read exactly what was said.
The four engines are not one thing. Perplexity leans on the live web, forums and news. ChatGPT draws on Bing's index and what it already knows. Gemini reads Google's index. Claude searches through Brave. A brand can be recommended by one and unknown to another, and a grid that shows that is a diagnosis: it tells you which source ecosystem you are missing from, not just that you are missing.
You pick the engines before you ask, and the button shows what the run costs. Perplexity and Gemini are free for your first three questions a day; ChatGPT and Claude run on credits, charged only when an engine actually answers. Once the free three are used, every engine runs on credits, so a paying user is never held up.
Reading the Grid
Example result, one question
QUESTION "best accountant for small businesses in Vancouver"
BRAND Acme Accounting DOMAIN acmeaccounting.ca
ENGINE MENTIONED HOW SOURCES CITED
Perplexity yes named in answer yelp.ca, cpabc.ca, reddit.com
Gemini no - clearlyca.com, cpabc.ca
ChatGPT yes your site cited acmeaccounting.ca, cpabc.ca
Claude no - clearlyca.com, bbb.org
WHAT TO DO WITH THIS
Write the page that answers this question -> Content Brief Generator
Get listed where AI reads: clearlyca.com, cpabc.ca
-> project competitors and prospects
Give AI a page to quote (named, not cited) -> Content Brief Generator
Track this question weekly -> create a project
Each engine gets a card: mentioned or not, how you were identified (named in the answer, or your own site cited, which is the stronger signal), the tone of the mention where there is one, whether browsing was on, and the sources the engine cited. The verbatim answer sits under the card. If a card says not mentioned but lists sources, those sources are the sites currently occupying the spot you want.
One answer is a sample. Ask the same engine twice and the wording, and sometimes the verdict, will differ. That is why the result says "mentioned in one of two engines" rather than giving you a score, and why the useful measurement is the same question asked every week.
What To Do With the Answer
The grid ends in a short list of actions built from what came back, each linked to the tool that does it.
Write the page that answers this question. The single biggest reason a brand is absent is that no page on its site answers the question in the words the customer used. Send the question straight to the Content Brief Generator, which comes back with the headings, entities and questions an answer engine expects to find, and draft it from there.
Get listed where AI reads. Where you were missing, the check shows what the engine cited instead. Those sites are the competitors that matter in AI search, and the places to be mentioned. Size them up in Competitor Intelligence, or create a project and they become tracked competitors and prospects for outreach.
Give AI a page to quote. Named in the answer but somebody else's page cited for it means the engine knows you exist and had nothing of yours worth linking. A page that answers the question directly is the fix, and the brief tool is where it starts.
Make sure AI can read the site. Absent from every engine usually has a boring cause: a robots rule blocking AI crawlers, or content that only exists after JavaScript runs. Every project includes a free AI Crawler Access check that fetches your site as each crawler does.
Track the question weekly. A project asks your set of questions across all four engines every week, keeps every answer, and tells you when a competitor moves in or the tone of a mention turns.
Questions
What should I ask?
The question a customer would type when they do not know you exist yet: "best [what you do] in [where]", "[category] for [situation]", "is [your brand] any good". Enter your domain and the tool suggests four, including the kind that does not contain your name, which is the one most businesses lose.
Why are only two engines free?
Each answer is a real call to that engine with web browsing on. Perplexity and Gemini are the two we can absorb, for three questions a day. ChatGPT and Claude run on credits, and so do the free two once your three are used. Credits are charged only when an engine actually answers, and they never expire.
Why did the engine not name me when I know I rank for this?
Ranking and being recommended are different. An engine writes one answer, names a few brands and links a few sources; it does not return a list. Check the sources it cited: that is where it looked, and it did not find you there.
Do the answers change between runs?
Yes. The engines are non-deterministic and their sources update. A repeat of the same question within a day returns the stored result, so a fresh reading means waiting, and a trend means a project.
Does this store my question?
The result is cached for a day so a repeat does not cost another engine call. Nothing is saved to an account unless you create a project and add the question to it.
One question tells you where you stand today and who is standing in your place. A project asks your whole question set, across all four engines, every week.
Free tools stay free · No card required