See why AI suggested something, and where it came from

Every AI suggestion in the studio is a proposal. Nothing changes until you choose Use, or the equivalent button on that card: Apply in AI Assist (for example Apply 2 changes), Use plan or Add pages to the course (for example Add 3 pages to the course) in Blueprint, and Build course in Draft with AI. Before you decide, the suggestion shows why it looks the way it does and what it drew on.

Read the rationale and sources

Each suggestion card shows:

• Why this suggestion: a short explanation, under 400 characters, of the choices the assistant made.
• Based on: the sources it relied on. A source can be the course’s Source text, a Framework level (for example bloom-revised:apply), a built-in Template, or one of your numbered Objectives.

The studio checks every framework and objective source against the course. If a source doesn’t match a real framework level or objective, it’s left out, and the rest of the suggestion is still shown.

You’ll see the rationale and sources on:

• AI Assist suggestions for a page or block.
• Draft content and Draft a question fixes for gaps in Learning Design → Alignment.
• Suggest level in Goals & objectives.
• Blueprint plans and new pages in Learning Design → Blueprint.
• The plan in Draft with AI.
• Draft with AI in the Rubrics panel.
• Selected-text rewrites in AI Assist and Write wrong answers on quiz questions.
• Review with AI on objectives, questions and rubrics. See Check an objective, question or rubric for bias, reading level and generic wording.

What the studio records when you use a suggestion

When you use a suggestion, the content it created or rewrote remembers its origin. This applies to objectives, framework levels, quiz questions, rubrics and blueprint pages. The record holds:

• that the content came from AI;
• which model wrote it, and which version of the studio’s instructions it followed;
• who chose Use, and when.

If you change that content later, the record notes that you edited it. In Draft with AI, objectives and pages you change in the plan before Build course are marked as edited from the start.

These records travel with course packages and duplicates. They’re reference information only: they never change how a course behaves for learners.

The decision log

For each suggestion the studio serves, it keeps a small log entry: the kind of suggestion, the model, the instruction version, and your decision: accepted, edited (you changed the suggestion before using it, for example a Draft with AI plan, a rewrite or wrong answers), rejected, or expired (the course changed before you chose Use). The entry also holds the IDs of what you used. It never stores course text. Only the course owner can see a course’s log.

Availability and limits

• The rationale comes from the assistant, so treat it as an explanation to check, not proof. Read the suggested content before you use it.
• Selected-text rewrites and wrong answers are logged, but they don’t add an AI record to the content: page text and individual answer options don’t carry one. A quiz question that already came from AI is marked as edited when you change its options.
• Agents that read a course see the same records and should send them back unchanged. See Apply operations safely.