Peer review

Reviewing a workflow against a rubric is where the course's ideas set. You'll spot the missing checkpoint or the vague prompt far faster in a design you're grading than in one you're defending. So the final act is two structured reviews of your capstone: one by you, wearing a grader's hat, and one by a colleague you recruit.

Round 1 — Review your own capstone

Budget about 30 minutes per review (the lesson minutes above count reading time; the reviews themselves happen across your week). Let the capstone sit for at least a day first, then read it as if a colleague had sent it to you:

1. 1Read the one-pager cold. Before opening the evidence pack, predict: where will this break? Write the prediction down.
2. 2Score the four rubric categories (25% each). Anchor every score to something specific — quote the prompt, point at the checkpoint.
3. 3Answer three questions in prose: What's the strongest design choice? What single change most improves it? Would you run this workflow with your name on the output — every week, including the busy ones?
4. 4Check the honesty axis hard: do the run logs support the headline numbers? Is review time counted? Generous scores for honest limits; hard scores for hype.

Round 2 — Recruit one colleague

Hand a colleague your one-pager, the rubric, and the same four steps above. They need zero AI experience — a skeptic is actually the better reviewer, because your design has to earn their trust the same way it will in a real rollout. Budget 30 minutes of their time, and offer to review something of theirs in return.

Acting on the reviews

Between your self-review and your colleague's, you now have two rubric-scored reviews. The rule: you must change one thing before the final version — the weakest point the two reviews agree on, or if they disagree, the one you know in your gut is right. Log the change in your one-pager's revision note. (Refusing to touch a reviewed draft is the human version of ignoring the checkpoint.)

The final package & the certificate

• Final one-pager (with revision note) + evidence pack + briefing script or slides.
• Both completed reviews — thoughtful reviewing is part of the skill this course builds, so keep them with the package.

Completing the capstone earns Certified AI Practitioner — Foundations (credential ID format EDOVA-AF-2026-XXXX). It certifies exactly what you've now done, not what you've watched: briefed a model precisely, built a checkpointed workflow, applied the responsibility rules, verified output like a professional, and taught it to someone else.