---
title: "Run AI verification of requirement content"
description: "Review the basis and rationale of each result rather than relying on a headline score."
url: https://helpcenter.complytrain.net/en/articles/147-run-ai-verification-of-requirement-content
collection: "Requirements"
updated: 2026-09-22
language: en
---

# Run AI verification of requirement content

Review the basis and rationale of each result rather than relying on a headline score.

Review the basis and rationale of each result rather than relying on a headline score.

## Start a run

Product location: `/reqms/verification`

Open **Requirements**, then **AI Verification** and choose the standard. Select **Run verification**. Your role needs **Assess Evidence** or **Manage Documents** permission.

A large run can require confirming the number of requirements and associated AI usage. Track the queued, running, completed or failed state and the number of requirements judged.

## Review the results

Product location: `/reqms/verification`

Read each verdict, source and rationale. **Fulfilled**, **Not fulfilled** and **No content** are different outcomes. A requirement with no mapped content cannot be treated as assessed merely because the overall run finished.

A result **Via equivalence** uses a cross-standard relationship and revalidates this standard's clause against the equivalent clause's content. Inspect that relationship and content when the result matters to a decision.

## Follow up changes

Product location: `/reqms/verification`

Open the underlying document or requirement to correct missing or inadequate content. Re-run verification after relevant changes and compare the applicable result.

The report can hide verdicts associated with retired equivalences while retaining them for audit. Check current scope and source state before comparing runs. An AI verdict supports review; it is not a human approval or an external certification decision.
