ASO AI assistant, in practice
A language model on its own is the wrong tool for keyword research: it will happily invent a search volume. What it is good at is deciding which of thirteen tools to run and reading the result back to you. That is the whole design of the assistant. It can find an app, read every phrase a competitor ranks for, compare two apps phrase by phrase, score one keyword, list who ranks beside an app, read a chart with movement, count ratings per country, compute a visibility score and tell you whether the store moved today. It cannot produce a number those tools did not return.
Each answer shows the steps it ran, with the numbers each returned, so you can see where a figure came from. Reports it starts are ordinary reports: they appear under Reports, cost one of the month's runs when a fresh one has to be made, and the phrases in them can be tracked. The assistant knows which apps are yours because you followed them, and if you have not followed one yet it asks for a name or a link and follows it for you in the conversation.
The same thirteen tools are exposed as an MCP server, so Claude Code, Claude Desktop, Cursor, Codex or VS Code can call them with one bearer token. Credits are metered: 40 a month free, 100 on Track, 2,000 on Pro. A message costs one credit plus one per tool it runs, so a quick question is 1 and a full competitor comparison is 4 or 5.
Written by Primo, who builds aso.best. We ship iOS apps ourselves; every method on this page is one we use on our own listings, and the hidden-field inference is tested against our own keyword field. Not affiliated with Apple.