1. Wire intake
The Apex Wire feed polls the publishers we follow on a schedule. Each run is logged with the number of items on the wire, how many were new, how many were duplicates of something we already had, how many were composed into a draft, and how many errors the run produced. That log is visible to the newsroom on the Wire Runs tab, so a bad run is obvious rather than silent.
Anything that matches a story we already hold is marked a duplicate and dropped. That is why you will not find the same wire item published three times under three different headlines.
2. Rewrite and expansion
A wire item is only useful if it can be rewritten without inventing anything. The rewrite produces a new headline, a standfirst that states what the source reported, a structured body that separates what is known from what is not, an explicit “why it matters” passage, and a short list of key facts extracted from the source material.
Where the source is thin, the rewrite says so. You will often see us write that the supplied account does not quantify something or does not establish something. That is deliberate: it is more honest to name the gap than to close it with a guess.
3. Originality check
Every story gets an originality score from 0 to 100, measured against the source text. The score is shown on the story page, next to the disclosure. A high score means the wording is substantially ours; a lower score means the rewrite stayed closer to the source. We do not publish a rewrite that is a light paraphrase of the source, because that is copying with extra steps.
The source's own headline and summary are kept in the database for the newsroom's originality review. They are never rendered on the public site as if they were our text.
4. Editorial review
Composed drafts go to a review queue, where a reviewer sees our rewrite next to the source headline and the originality score. The reviewer can approve, edit, block or re-queue the story for another rewrite. Only stories that clear review are published. Raw, needs-review, duplicate and blocked stories are reachable only from the newsroom console.
5. Publication and attribution
On publication, each story receives a permanent link, an attribution block at the top of the body naming the outlet that first reported it and the time it appeared on the wire, and an outbound link to the original markedrel="nofollow noopener". The attribution block is part of the article, not a collapsible footnote, and it is never removed, truncated or hidden.
Images and covers
We never hotlink or scrape an image from another publisher, and we never invent photography or depict a real person's likeness. If a story carries its own supplied Hero Image, we use it and say so. Otherwise we generate a branded cover from the desk palette, the category and the story's leading entity. The cover is deterministic: the same story always renders the same image.
Automation and AI
Drafting and fact extraction are machine-assisted; the decision to publish is human. Every story that reaches the site has been approved by a reader with an editorial role. The wire-run log records how many drafts a run produced and how many were actually published, so you can see the human filter working.
What we never do
- Publish a story without naming and linking the original outlet.
- Present a source's headline or summary as our own reporting.
- Show a real person's likeness or a stock photograph as if it were the subject of a story.
- Invent a number, a quote, a source or a reaction count. Aggregate figures come from stored fields and rollups.
- Remove or soften an attribution block after publication.
Corrections
When we get something wrong we fix the story and record the change. See the corrections page for how to request one and what happens next.