An independent index of what AI assistants recommend

I built validcube to answer one question for myself. When people stop browsing the App Store and start asking ChatGPT, Claude or Gemini "what's the best app for X," which apps come back? Those answers are quietly becoming the shelf, and almost nobody is tracking them in the open.

This is a research project, not a product I am selling. Once a month I run a fixed set of buyer questions across five assistants and record which apps each one names. A single month is a photo. The point is the film: how the answers move as models get updated and as apps gain or lose ground on the web.

What it is honest about

The numbers come from APIs on the fast, cheap model tier, not from the exact logged-in app with your memory and history behind it. That gap is real, and the methodology says so plainly, including the exact model IDs and the scoring formula. The value here is not precision. It is that the same questions run the same way every month, in public, by someone with nothing to sell.

Who runs it

validcube is a project by Wolfgang Peters. Contact details are in the Imprint; how your data is handled is in the Privacy Policy.