When someone asks an AI assistant which app to download, we count how often it says yours. 45 asks per engine, every week. Monday morning you get the card.
Every AI-visibility tool out there measures a domain and hands you a score out of 100. We measure the listing people actually install, and we count.
Not a dashboard to check. A card that says who got named instead, what the engines read, and the three listing changes most likely to move the count.
Ask the same question twice and you get two different answers. So we never say “you are #3”. We say 7 of 45, likely 8–29%, and we tell you when a change is real and when it is noise.
I built AppCited because I wanted to know whether ChatGPT ever recommends Fitvibe, my own app, and nothing could tell me. Now I am building the tool that tells me what to fix, in public, with the Launchspace community. The first Monday cards go out to this list.
Ask the same question on two days and you get two answers. Daily trackers show you that noise as movement. We ask 45 times a week instead, so the count comes with a range, and we only call a change real when the ranges stop overlapping.
Close, not identical. We use the official APIs with web search on, so results are repeatable and labelled “API panel”. The methodology page shows how we check it against the consumer apps.
Most indie apps start at 0 of 45. The card shows who got named instead and what the engines read, so you know where to show up first.