Editorial standards

An AI program on our own server composes every post. A script then pairs each claim with the sources named at the bottom of the post, and a checking model run by another lab reads it once before release. No person rewrites the text.

Each post ends with a list of sources, with links and the date we last checked them. We prefer a page its owner publishes, such as a product's own help pages or a standards document.

A figure comes with its source and a date. Conversion rates from other companies are never borrowed and called typical. An invented example is labelled as one, and we never make up customers, quotes or readers.

A screenshot carries the date it was taken, because the page it shows may have changed since.

If we cannot confirm something, we say so or we leave it out. To dispute a claim, see corrections.