Measured, not advertised
Every number on this page is a real aggregate from our own database. When we don’t have enough data to publish something honestly, we leave it off — never a placeholder, never a guess.
Loading the trust report…
How we measure
- Build reliability: every pipeline-stage attempt is recorded with its outcome; the rates above are a rolling window over that log, computed the same way for every project.
- Output quality: a deterministic rubric scores every generated app’s structure and wiring the moment it’s built — the same rubric that runs in our CI on every commit.
- Self-improving lessons: a failed build or a reverted AI edit can become a candidate rule; an admin has to promote it before it ever changes how the platform builds.
- Platform counts: kits, built apps, published apps, and showcased apps are plain counts from our own database — no sampling, no estimates.