TL;DR
- Hi, Moose chat has a new instant check: ask "Does AI recommend my business?" and it queries ChatGPT, Perplexity, Grok, Google AI Mode, and Google's AI Overviews in parallel, from your own computer, for free.
- The check uses an unbranded question a real customer would ask, then shows per engine whether your business was mentioned and whether your own site was cited.
- With the local model loaded, or your own OpenRouter key, each answer also gets a verdict: recommended, listed among others, described neutrally, or described negatively.
- One click adds the question to your tracked prompts, so a one-off snapshot can become a daily trend.
- Also this week: a Tools menu in the top bar, AEO audits that read your recent visibility data, and visibility reports that no longer count failed runs.
The first question
Everyone who hears what Hi, Moose does asks the same thing within about ten seconds: "so, does ChatGPT recommend my business?"
Until this week the honest answer was "set up tracked prompts and wait for the first scheduled run." Which is the right answer for building a trend line and a terrible answer for a person who is curious right now. People do not arrive curious about methodology. They arrive curious about themselves.
So now you can just ask.
One question, five engines
Type the question into chat and Moose sets up the check. It frames an unbranded question the way a customer would ask it, with no brand name in it. That part matters: what AI says when someone searches your name is mildly interesting, but whether you come up when nobody hands the engine your name is the whole game.
Then it queries five engines in parallel: ChatGPT, Perplexity, Grok, Google AI Mode, and Google's AI Overviews. The fetches run from your own computer, the way a browser would load those pages, so the check is free and uses none of your plan allowance. It takes a minute or two. Grok likes to think.
As each engine answers, a card fills in: mentioned or not, your own site cited or not, and the sources the answer leaned on. If you have the local model loaded, or your own OpenRouter key, each answer also gets a verdict on how it treats you: recommended, listed among others, described neutrally, or described negatively.
One run is one snapshot. AI answers wobble from day to day, which is why the result card has a button to add the question to your tracked prompts, where it runs on a schedule and builds the trend.
A menu for the tools
The chat could already do a lot of things, provided you knew to ask for them. That is a bad system. So the top bar grew a Tools menu that lays everything out in three groups: audit and capture, enrich a page, and write. AEO audits, ChatGPT fan-out capture, grounding queries, FAQs, key points, audio versions, content drafts. No memorizing, no guessing at magic words.
Audits that have seen your data
An AEO audit tells you how ready a page is for answer engines. Visibility tracking tells you what those engines say about you today. Those two lived side by side without talking to each other, which always felt slightly silly.
Now the audit pulls in your recent visibility data when it exists. If Perplexity already cites one of your pages, the audit knows, and its recommendations lean on that evidence instead of generic best practice.
The boring load-bearing part
Visibility runs also got a round of efficiency and reliability work, and a run that fails partway no longer leaks into your reports. Nobody will ever notice this, which is the point. A trend line is only worth reading if the bad data points were never allowed in.
FAQ
What does the new AI brand check do?
It asks one unbranded, customer-style question across ChatGPT, Perplexity, Grok, Google AI Mode, and Google's AI Overviews, then shows per engine whether your business was mentioned and whether your own site was cited. It runs from your own computer and finishes in a minute or two.
Does the instant check cost anything?
No. The engine fetches run locally from your machine, so the check is free and uses none of your plan allowance. The optional grading of each answer runs on the local model or your own OpenRouter key.
Why does the check use an unbranded question?
Because a customer who does not know you exist will not type your name. The useful signal is whether AI engines bring you up on their own when someone asks the kind of question your customers ask.
How is this different from tracked prompts?
The instant check is a one-off snapshot, and AI answers vary from day to day. Tracked prompts run the same question on a schedule so you get a trend instead of a single reading.
The one who's under the desk for all of it.
