Verity

Verity — evidence before publication

Verity is a claim-review workspace for journalists, researchers, and teams publishing consequential information. Instead of returning a black-box answer, it keeps the claim, verdict, evidence, source links, provider context, and confidence in the same review surface.

The Build Week focus

This submission centers on one practical workflow:

Paste a high-stakes claim.
Run a source-backed verification.
Separate what the evidence supports from language that is too broad, stale, or misleading.
Save the result and export a review-ready record.

The goal is not to replace editorial, legal, or medical judgment. It is to make the evidence trail inspectable before a claim is published.

Desktop workflow

The desktop app supports Quick, Standard, and Deep verification modes; sends a claim to Verity's verification API; renders the returned verdict, confidence, sources, and provider context; saves results to history and a workspace; and can export a verification report. If the API is unavailable, its local fallback is explicitly marked as provisional rather than presented as a verified result.

Why it matters

Fast publishing workflows are vulnerable to confident wording, stale citations, and claims that overgeneralize a narrow study result. Verity turns those issues into a reviewable evidence record, so a team can correct the wording before the claim reaches an audience.

Build Week integrity

This draft deliberately avoids unsupported public accuracy, compliance-certification, latency, and provider-count claims. The demo will show a real evidence-backed verification flow and will clearly identify the role of Codex and GPT-5.6 in the submitted build.