Automated accuracy, tone, accessibility, and completeness checks, plus human approval before anything reaches learners.

Review gates are the checks a course must pass before it is published. They combine automated checks with a human sign-off, so AI speeds up authoring without lowering the bar.

The automated checks

Check	What it looks for
Accuracy	Claims that are not supported by, or contradict, the attached sources. Every flagged claim links to the closest source passage.
Tone	Passages that drift from your brand voice, glossary violations, and banned terms.
Accessibility	Missing alt text, media without captions or transcripts, low color contrast, and content that is not keyboard operable.
Completeness	Learning objectives without matching content or assessment, empty sections, and broken links.

Each flag shows the passage, the reason, and a suggested fix. Accept the suggestion, edit manually, or dismiss the flag with a note.

Human approval

After the automated checks, an approver signs off. Approvers are typically the course owner or a subject-matter expert; for regulated content, often someone from compliance or legal.

Nothing ships unapproved

A course cannot be published until every blocking flag is resolved and an approver has signed off. Dismissed flags and approvals are recorded in the audit trail with who, when, and why.

Typical review workflow

Author submits for review

The author runs the checks and resolves what they can.

Reviewer inspects flags and content

The reviewer reads the course, checks remaining flags, and leaves comments on specific blocks.

Approve or request changes

Requested changes go back to the author. Approval publishes the version (or schedules it).

When you edit a published course, the changed blocks are re-checked and re-approved. You do not have to review the whole course again for a one-paragraph change, but the new version is recorded separately so compliance records show exactly which version each person completed.

Review gates