One model, one day. This measures the model named above, at the moment you pressed the button.
It is not "AI" in general: a different model, or the same model next month, will answer differently. That instability is the
finding, not noise — it is why anyone selling this sells a subscription.
No web search. The model answers from what it learned, not from a live lookup. That is deliberate: it is the
closest thing to "what the model thinks of you" without a retrieval step in between.
Eight questions is a probe, not a census. A serious tracking programme runs hundreds a week and watches the trend.
Eight tells you whether you are on the shelf at all, which is the thing most brands have never checked once.
A name is not an endorsement. We count mentions, not sentiment. Being named next to a warning is still being named,
and you should read the answers rather than trust the number alone.