Who We Are

The short answer

We could have written an invented team page with invented titles and invented biographies. We did not. There is no company, no office and no staff behind this site; there is an operator making the decisions and AI agent sessions doing the work.

Writing this down is not a display of modesty but a matter of consistency. A site that tells you its form does not work cannot then present a team it does not have.

How the work runs

Every round starts with a written analysis: which pages, where the data will live, which routes, which questions are still open. No code is written before that analysis is approved. The reason is simple: scaffolding is cheap to produce and a wrong data schema is expensive to undo.

When production is finished, a single verification command runs. It checks more than the tests: it also checks real page responses against a server that is actually up — because once in this repository all four gates were green while the page rendered completely unstyled, and not one gate said so.

A step that could not be run is not skipped; it is reported as "could not run". When something goes wrong, a one-line record is opened and a mechanical measure is sought that makes the same mistake impossible to repeat.

How that shows up on the site

Some of the choices you will see here are unusual: a form that does not work says so, an empty detail row is not printed at all, and a filter button that would return nothing is not shown.

All of it follows from one principle: the most expensive failure in an interface is not crashing, it is appearing to work.