NEWS

Curated, verified, essential.

Wrong password

Secure Editorial Access
INTELLIGENCE

NEWS

search

Featured Briefing

No articles yet

Methodology Explainer

Curated by Editorial Intelligence AI

Methodology

How This
Works

An AI-curated news briefing built on transparent source selection, structured analysis, and editorial priorities you can inspect.

verified

Source Registry

We pull from 30 curated RSS feeds spanning 11 countries and regions. Sources are tiered by editorial standards, correction history, and independence. The AI editor is instructed to prefer higher-tier outlets and cap any single source at 3 articles per edition.

Tier 1: Primary Sources Wire services, public broadcasters, papers of record
Reuters · AP · BBC · NHK World · ABC Australia · Nikkei Asia · Japan Times · AFR · The Economist · Financial Times · DW News · Channel News Asia · Lowy Institute
Tier 2: Specialist & Analysis Domain expertise, strong editorial standards
NYT · WSJ · The Guardian · Bloomberg · Al Jazeera · Asahi Shimbun · Mainichi Shimbun · Yomiuri Shimbun · MIT Technology Review · TechCrunch · The Verge · Ars Technica · France 24 · Straits Times · Project Syndicate
Tier 3: Niche & Community Valuable for specific topics, used selectively
Disrupting Japan · AJBCC
edit_note

Curation Process

Each edition is assembled in five phases. Two use AI (Claude Sonnet at high effort), three are deterministic code (RSS ingestion, image sourcing, publishing).

1. Selection

An AI editor reviews all available headlines with source trust tiers and picks 25-30 articles per edition. Codified editorial priorities: breaking news first, then regional relevance (Japan, Australia, Asia-Pacific), AI/tech, business, cybersecurity, science, then global affairs. Featured stories are chosen for impact and category diversity. Listicles, sponsored content, and thin rewrites are filtered out.

2. Analysis

Each selected article is fetched and read in full by a sandboxed AI analyst. It produces a rewritten headline, summary, and detailed account with structured metadata: bias score, verification status, claim count, and fallacy flags. Original reporting and analysis are prioritized over wire summaries.

balance

Bias & Verification

Bias Mapping

Each article receives a bias score (-3 to +3) and directional label. This isn't about removing perspective — it's about making it visible so you can read with awareness.

Claim Tracking

The AI counts factual claims and assesses how many are verifiable from the source text. Articles are marked verified, partially-verified, or unverified based on this assessment.

psychology_alt

Fallacy Detection

Articles are scanned for common logical fallacies: Strawman, Ad Hominem, Slippery Slope, Whataboutism, Circular Reasoning, False Dichotomy, and Appeal to Authority. Flagged fallacies appear in the article metadata.

warning
Transparency Note

This analysis is performed by AI and reflects its interpretation. It's a reading aid, not a fact-check. Always read critically.

verified Pipeline v7.0 · 30 sources · 11 countries