What it does
Measures whether the assistants your customers actually ask are aware your product exists. It puts the same buying questions to more than one model, records what each recommends, and reports where your brand appears, where a competitor appears instead, and what the models believe about you that is out of date or wrong. Search rankings stopped being the whole picture the moment people started asking a chat window "what should I use for X". That surface has no ranking report, no console, and no way to appeal — the only way to know where you stand is to ask it and write down the answer. The discipline is in treating each run as a sample, not a measurement. Model answers vary between runs, between versions, and between phrasings of the same question. So it asks each question more than once, reports how consistent the answers were, and refuses to call a single mention a trend. A one-off appearance reported as a win is how this kind of tracking becomes theatre. What it finds that nothing else does: the specific wrong belief. "Three of four runs said you don't have an API" is an actionable content gap in a way that a visibility percentage never is.
Example prompts
- Ask both models 'best tool for X' four times each and tell me where we appear.
- What do the models believe about our product that isn't true any more?
- Which competitor comes up when we don't?
Before you connect
The credentials this agent will ask you for — the full setup is on the Setup tab.
Needs 2 credentials (1 required) to connect. See setup
