Questions, review, and AI suggestions

Writing questions, approving them before they go live, and handling suggestions.

5 min read

Writing questions

Content → Authoring is where questions live, per course. A good question tests whether someone can act correctly, not whether they memorised a phrase — prefer "what do you do when…" over "what does the SOP say about…".

• Make the wrong options genuinely plausible. An obviously silly option teaches nothing and inflates scores.
• Write coaching for each wrong answer, so picking it explains the mistake.
• Mark a question safety-critical when getting it wrong could hurt an animal or a person. Those are weighted accordingly.

Nothing goes live unreviewed

Content → Needs review holds questions waiting on a human. This is the gate: a question only reaches learners once someone approves it. If your team is seeing odd questions, this queue is where to look first.

AI suggestions

Content → AI suggestions holds proposed additions to content you already have — coaching for wrong answers, safety warnings on critical lessons, images. Nothing here is live.

1. Use the tabs across the top to review one kind at a time — safety warnings are a different job from answer coaching.
2. Read the suggestion. If it's right as written, press "Approve as written".
3. If it's close but the wording is off, press "Edit…", fix it, then "Publish edited".
4. If it's wrong, Reject it.

Checking for duplicates

Content → Overlap check finds content that covers the same ground twice. Worth a look after importing a new document, since SOPs repeat each other constantly.