AI Mention Finder

AI mention finder

Ask ChatGPT, Claude, Gemini and Perplexity the questions your buyers ask, then see who gets named, in what order and in what tone. Up to three competitors side by side, every answer shown in full.

Competitors to compare (up to 3)
Prompts to ask (0/3)

Unbranded questions show whether the models bring you up on their own. That is the real visibility test.

Models

4 answers per check ยท 3 checks per 10 minutes

Nothing tuned in yet. Fill in your brand and a category, add one to three prompts, then run the check. Each model answers each prompt once, right now, and every answer is shown in full so you can read it for yourself.

What one check actually tells you

Each check is a fresh, live sample. The tool sends every prompt to every model you picked, once, with no chat history. It then scans each answer for the exact names you entered and records three things:

  • Whether you were named at all. The simplest and most honest signal.
  • Your order among the names you entered. In a list of recommendations, first place is read first.
  • The tone around the mention. Being named as "a pricier option with a steep learning curve" is not the same as being the top pick.

What it does not tell you: how many people ask these questions, what any individual user saw, or your share of a whole market. Nobody outside the AI companies has that data, and tools that print a precise "AI share of voice" from a handful of prompts are describing their sample, not the market. Our methodology page spells out the limits.

How to use it well

  1. Start unbranded. Use the best-of, recommendation and alternatives templates. Branded prompts only tell you what a model believes about you once asked.
  2. Use the words buyers use. "Invoicing app for freelancers" and "accounts receivable software" are different conversations with different winners.
  3. Repeat before you react. Language models vary between runs. A brand that appears in 1 of 4 answers today may appear in 3 of 4 tomorrow. The AI visibility guide has a slider that shows how wide that uncertainty really is.
  4. Read the answers. The reasons a model gives for recommending a competitor are your to-do list: missing comparisons, pricing pages it can't find, reviews it cites.

Questions

How do I check if ChatGPT mentions my brand?
Ask it the questions your buyers ask, without naming yourself, and see whether you come up. This tool does that for you: it sends up to three prompts to up to four models (an OpenAI GPT model, Claude, Gemini and Perplexity Sonar) through their APIs, finds every exact mention of your brand and up to three competitors, and shows each full answer.
Is this the same as what people see in the ChatGPT app?
Not exactly. The check calls the models through their APIs with no chat history, no personal memory and no location. Consumer apps add their own system prompts, memory and sometimes web search, so answers can differ. Treat the result as a clean, repeatable sample of how the underlying model answers.
Is the percentage my AI market share?
No. "Named in 4 of 6 answers" is exactly that: a count from this sample, at this moment. It says nothing about how many real people ask these questions, and a single run can swing noticeably. Run the same prompts again on different days and compare saved checks before drawing conclusions.
How is sentiment worked out?
After the answers arrive, a separate GPT-6 Luna call reads only the text around each mention and labels it positive, neutral, negative or mixed with a short reason. You can open every answer to check the label yourself.
Does it cost anything or need an account?
No. There is no sign-up. To keep it free, each visitor can run 3 checks every 10 minutes and 10 per day. Saved checks stay in your own browser.
Why do unbranded prompts matter most?
If you ask "What do people say about Acme?" every model will talk about Acme. The useful question is whether a model brings you up when a buyer asks for "the best invoicing app for freelancers" and never types your name. That is what AI visibility means in practice.