Picks the question for you
Generated from your product context so the first run needs no setup. The question is shown verbatim, because a competitor list you cannot trace back to a question is just a list of companies.
Monitoring
Your market, as an assistant describes it, usually without you in it
1
buyer question, asked on your behalf, before you configure anything
The first-run baseline in internal/activation. It ends onboarding on a finding rather than on a form.
Onboarding used to end on configuration, which meant the first thing a paying customer saw was an empty inbox. It ends here instead. We take the question your buyers ask, put it to an assistant, and show you the names that come back, ranked by how often they were mentioned.
Both "mentioned" and "cited" being false is the common outcome on day zero, and it is also the most motivating one. It defines the game in one screen: get into that answer.
Where it lives: The last step of onboarding, and revisitable from the dashboard afterwards.
What it does
Names who got recommended
Competitors are read out of the answers themselves, most-mentioned first, whether or not they are in your tracked competitor set. Some of the more useful findings are the names a founder had never heard of.
Re-runs when you reposition
Changing your positioning invalidates the baseline and the mirror ignores checks older than that reset, so it can never show you the names your previous positioning collected.
How it works
- The baseline is claimed atomically, so two open tabs cannot both spend a set of AI calls.
- Time from starting the baseline to seeing the finished mirror is measured, because time-to-first-insight is the number that decides whether a new account survives its first week.
- A failed baseline is recorded as failed and retried on the next visit rather than left spinning.
What it does not do
- One question on one engine is a sample of one. It is a starting pistol, not a measurement.
- Competitor extraction reads names out of prose, so an answer that describes a category without naming vendors gives you an empty list.
- Seeing a competitor here does not mean they are winning generally. It means they were in that answer, that day.
Which plans have it
Runs once per product on every plan, and again after a repositioning.
Read the full plan comparison.
Frequently asked questions
How long does the first run take?
A generation and a check, so under a minute in normal conditions. If a provider is slow the screen tells you it is still running rather than pretending to be finished.
What if I am already mentioned?
Then the interesting question moves to whether you were cited, and to which pages the answer was built from. That is what Citation Sources is for.
Does it use my competitor list?
It reports whoever the answer named. Your tracked competitors are used for share-of-voice elsewhere, but the mirror deliberately does not filter to a list you wrote, because the names you did not think of are the point.
Related features
Monitoring
Every feature we publish AI visibility monitoring
Scheduled per-engine checks with a 0-100 score, tracked separately for being named and for being linked.
MonitoringPrompt library
Auto-generated buyer questions, editable by hand, regenerated when your positioning changes, rotated by staleness.
Citation corpusCitation Sources
Every URL the engines cited in your category, classified, checked for whether you appear in it, and sorted by what you can actually win.
Run it on your own product
Start with the free visibility check to see whether the engines name you today, then run the scan that shows which pages they used instead.