Editorial standards
All We Need is produced by an automated editorial system with explicit source, research, model-routing, quality, and publication gates. These standards describe what the system is allowed to publish—and what must remain deferred.
Official releases, filings, research papers, transcripts, and direct public statements are preferred. Reputable reporting can establish or corroborate an event. Aggregators and discovery alerts can initiate research but cannot substitute for citeable evidence.
Unattributed or indirect signals must be traced to accessible originals. Material stories are checked across independent sources when available. Evidence roles and takeaways remain attached to the article.
Task-specific models assist with grounding, editorial research, analysis, writing, and localization. Model output is never treated as evidence. Schema, completeness, source coverage, and quality checks must pass before publication.
Focus articles distinguish reported facts from editorial synthesis. Explore articles are explicitly thesis-driven and must include uncertainty, disconfirming evidence, or a counterpoint.
Research failures, incomplete localization, source-health collapse, old-story protection failures, test failures, and deployment verification failures stop publication. The previous verified site remains online.
New evidence may be appended to an existing story without silently rewriting its history. Material factual errors should be corrected promptly and identified in the article’s update trail. Questions or correction requests can be filed through the public issue tracker.
Found a factual error, broken source, or misleading synthesis?