A bespoke content engine that takes a topic and produces a publish-ready, fully-cited authority article — engineered for AI search (GEO) and held to a strict editorial QA gate — in under an hour instead of the two to three days it used to take.
The engine builds a content roadmap from what the audience is actually asking, then researches UK primary sources — government departments, regulators, peer-reviewed papers, named industry surveys — pulling five to ten citable datapoints. Every figure carries a specific number, unit, date and source URL; anything failing those four tests is dropped.
Articles are written to the client's editorial template: a 40–80 word definitive answer paragraph at the top (the part LLMs lift verbatim), question-format headings that mirror real user prompts, at least one comparison table, structured metadata for AI crawlers, and an original calculation derived from public data where one is available.
A 10-element editorial gate checks citation density, comparison table, structured metadata, author byline and review date against the client's published standards. A separate voice scan blocks the article if it contains any of roughly 40 patterns that signal AI-written prose. Nothing ships until every gate passes.
The engine runs on a schedule — two articles a week plus a fortnightly fast-react slot for breaking news — saving each draft, styled preview, sources file and QA scoresheet to a private preview library. It then emails the client a preview link with a one-click approve or revise reply, so nothing reaches the live site without sign-off.
