Trust, documented

Editorial policy

A repeatable standard for humans and AI agents contributing to Index Us.

Source hierarchy

Use primary sources first: vendor documentation, model cards, research papers, repositories, regulatory text and direct statements. Secondary reporting can add context but should not be the sole evidence for a material claim.

Testing and recommendations

State what was tested, when, on which plan or version, and what was not tested. A product announcement is not a product review. Commercial relationships, referral links and supplied access must be disclosed prominently.

Automated reporting

Index Us uses an automated workflow to research, draft, edit and publish AI coverage. Articles produced through this workflow carry an explicit disclosure beside the byline. They are published under Index Us Editorial and are not reviewed by a human before publication.

Each article must pass a separate writing edit and a separate AI review of its claims against the linked sources. We require primary evidence and independently produced context, distinguish company claims from demonstrated results, and retain a record of the checks. These checks can miss errors. Clear prose and a passing automated review do not guarantee accuracy.

Selection and originality

Research runs are opportunities to find useful stories. We skip thin announcements, duplicates and stories without adequate evidence. Long-form coverage must add original explanation, context and practical value. We do not invent interviews, tests or personal experience, and we do not attribute automated articles to a human author.

Corrections

Material corrections should be dated and explained at the end of an article. Minor typography fixes need not be logged. Quietly changing the substance of a claim is not acceptable.

Freshness

Every article includes published and last-reviewed dates. Time-sensitive instructions should be rechecked against current documentation before republishing or recommending them.