Coral Wire is a daily AI-news wire for authors, published by Coral Hart Group. We track AI, publishing, and the book trade and cut it down to what actually matters for working writers. coralwire.ai
Coral Wire aims to publish a new edition every weekday morning: the AI, legal, platform and market stories that moved the author business in the last 24 hours, each one graded for what it actually means to a working author. The archive is the record of which mornings that happened — a missed edition shows up there as a gap, not hidden behind a blanket claim. Alongside each edition we publish two numbers of our own — The Coral Index, a 0-100 reading of how the day landed for authors, and the AI Tool Leaderboard, a count of which platforms authors are hearing about. Both are free to cite.
Found with experience · Edited by Coral Hart. Coral Wire publishes two different kinds of page, made two different ways. The wire listing — the scored headlines on the homepage, in The Wire: Latest and in the archive — is found and graded by an automated pipeline: it currently reads a fixed set of publisher feeds, industry press and court dockets, scores each story for what it means to a working author, and an Anthropic Claude model writes the headline and one-line summary directly from the linked source. When the pipeline surfaces a story through an aggregator, it resolves the link back to the original outlet before that outlet is cited — Coral Wire does not cite the aggregator itself. The written stories — the longer story pages carrying a byline and, where written, a “Coral’s read” — are reported, checked against their original source and worded by Coral Hart, who is accountable for what runs. Every written story has been read by a person before publication; the wire listing’s machine-written headlines have not — that is the honest difference between the two.
Coral Wire selects stories by a Claude-graded relevance score and writes each headline and one-line summary with an Anthropic Claude model from the linked source. Story gathering is not AI: it is a fixed set of feeds and dockets. The linked original is always the authority. Stories are gathered from a fixed set of publisher feeds, industry press and court dockets. An Anthropic Claude model grades each one for author relevance, sentiment and urgency; the highest-scoring stories become the edition, and a Claude model writes the one-line chyron and summary you read on the site. A human sets the editorial policy, the watchlists and the scoring rules, and can override the day’s index. We do not paraphrase a source we have not linked, and every story page carries the original URL.
The wire reads publishing trade press, law and policy primary sources, retail-platform announcements, and AI companies’ own newsrooms. Forums and social posts are used only as tip-offs — checked against a primary source before they inform a story — and are never themselves cited. We never cite our own group’s sites — PlotProse, TropeSmith, Perfect Year HQ, Book Ad Pack or Coral Hart Group — as news.
There is a Spot something wrong? form at the foot of every story. It files straight to the desk, tied to the story it is about, and you do not need to leave your name. We check the report against the story’s original source; if we got it wrong we correct the page and say so on the page itself, dated, rather than editing it quietly — see the update notes on our 22 September settlement stories for what that looks like. Leave an email if you want to be told what we found. You can also write to info@coralwire.ai.
Coral Wire is published by The Coral Hart Group, which also builds PlotProse, TropeSmith, Perfect Year HQ and Book Ad Pack for working authors. That is a commercial interest in the author-software market, and it is why the wire links its sources: check them.
RSS · JSON feed · Archive · Daily email sign-up on the front page
Every story here links its sources so you can check them. If we have got a fact, a figure or a date wrong, tell us and we will correct the page and say so on it.
Coral Wire selects stories by a Claude-graded relevance score and writes each headline and one-line summary with an Anthropic Claude model from the linked source. Story gathering is not AI: it is a fixed set of feeds and dockets. The linked original is always the authority.