How every output is scored

Made-up details

We list the details each tool made up that the brief didn't give: names, places, prices, policies, opening hours. It's there for reference and never scored. Same for the one-line verdict on each card.

Time to build

Wall-clock time from submitting the brief to the first complete, viewable page, timed the same way for every tool. If a tool kept working after that, the run page shows when it finished as a lighter “done” marker.

Reference only. Never scored.

One prompt isn't a benchmark

Fair. One run is one brief, one attempt per tool, on one day. It shows what each tool made that time, not everything it can do.

The brief and the method are both versioned, so runs line up over time. The picture gets better every month.

1. Method v1.0 · Oct 2026

   • First published method: five weighted criteria scored 1 to 5, overall out of 100.
   • Time to build is measured the same way for every tool, from submit to the first complete, viewable page. It is shown for reference and never scored.
   • Made-up details are listed for reference and never scored.

Briefs are versioned separately (Run 001 uses brief v1.0). A brief only changes between runs, and the run page lists what changed.

Affiliate links and sponsorships: read the disclosure.