Keldura Daily · Transparency

How Keldura Daily is made

Keldura Daily is a daily desk of short, source-linked briefings across a fixed set of topics. Each briefing distills what independent sources reported, links back to them, and leads with why it matters.

How stories are found

Every topic has a curated set of sources — major wires plus specialist outlets — and an additional open news search that catches the day's story when the curated sources miss it. Sources are ingested through the same provider pipeline the Keldura product uses: web articles, RSS feeds, and, where enabled, video, preprints, and social posts. Only recent items inside the edition's date window are considered.

How briefings are written

An AI model synthesizes the discovered sources into a short briefing — a headline, a one-paragraph summary, a "why it matters" line, and a compact cheatsheet. Every claim is tied to a numbered citation that links back to an original source.

How they're checked

Before anything is published, each briefing passes automated grounding review: hard gates — a minimum number of cited sources, English-only verification, and safe outbound links — plus an automated quality judge that scores how well the briefing is grounded in its sources. Briefings that don't pass are not published.

How they're published

Editions publish automatically once they pass review. On a day when a topic has no genuinely new development, no edition is minted rather than filling space — so "today's desk" stays literally true.

What we don't do

• No AI is credited as an author.
• We don't machine-translate the Daily.
• We don't re-date old briefings to make them look fresh — a briefing's date is the day it was published.