For Journalists and Fact-Checkers

What Veridi adds to your workflow

You already know how to evaluate claims. Veridi doesn’t replace that expertise; it provides a structured framework that makes the evaluation process explicit, repeatable, and auditable.

Three aspects are most relevant to professional fact-checkers:

  1. A formalized source hierarchy with structural confidence caps that prevent volume from substituting for quality.
  2. Explicit gaming countermeasures: named detection procedures for eleven disinformation techniques, applied systematically to every assessment.
  3. Decision trees for the hard calls: documented logic for resolving the distinctions that experienced fact-checkers often handle by instinct: Misleading vs. Lacks Context, Mixed vs. Mostly False, and similar boundary cases.

How it’s implemented

Veridi is a methodology - a set of documented procedures, decision trees, and evaluation frameworks. It’s currently implemented as a prompt system for AI (Claude by Anthropic), which follows the methodology to produce assessments.

This means:

  • The methodology is separable from the AI. The procedures could be followed by a human fact-checker, adapted for a different AI system, or used as a training framework.
  • The AI doesn’t decide what’s true. It follows the documented process - source hierarchy, decision trees, gaming countermeasures, confidence calibration - and produces an assessment that shows its reasoning.
  • When the AI makes an error, the methodology provides a framework for identifying where the process broke down and correcting the error prior to returning a result.

What this is not

Veridi is not a competitor to PolitiFact, Snopes, AFP Fact Check, or any IFCN-affiliated organization. Those organizations do essential work. Veridi’s contribution is a structured, auditable process that can complement existing practices, providing a second-opinion framework, a training tool, or a systematic check on the most common disinformation techniques.

Next steps