How it works
Six steps. The AI is only allowed into four of them.
Every mission runs the same loop. The stages are enforced by the platform, not suggested in a teacher guide — a student cannot skip to the answer, and the AI cannot skip to it either.
- 01 curiosity
Think
The AI is switched off here. Work out what you already know and what’s missing.
- 02 clarity
Ask
Send your first request. It won’t be your last.
- 03 creativity
Explore
Push past the first answer. Ask for other approaches.
- 04 critique
Challenge
Find what’s wrong, missing, or assumed.
- 05 choice
Decide
The AI stops here. This one is yours.
- 06 choice
Explain
Say why. In your own words.
The part that matters
The AI is deliberately less helpful than it could be
A perfectly helpful assistant ruins the lesson. If a student writes “plan a pizza party” and gets a flawless plan back, the student has learned that vague questions work.
So the C5 coach answers what was actually asked. Vague in, generic out — plus one question pointing at what it would need to do better. It never hands over a “better prompt,” because the goal is not better prompt writers.
At the decide stage it stops answering entirely. If a student asks it to pick, it declines once, briefly, and offers to lay out the tradeoffs instead.
How it’s scored
Two layers. First, behavior counted without a model: did the student supply constraints, ask a follow-up, seek evidence, reach a decision, justify it. Those counts cap the rubric — a one-message session cannot score Advanced at Critique no matter how it reads.
Second, a rubric evaluation against the C5 levels that returns the quotes it based each level on, so a teacher can see the evidence without reading the whole transcript.
Nothing rewards message count, message length, or time spent. A short exchange that surfaces a hidden constraint outscores a long one that doesn’t.
Try the decision
Sample mission · The Pizza Party
Your class has $250 for a pizza party for 40 students. The AI has already made its recommendation.
Mark a box. Then find out what nobody asked.
What comes back
The read-out
Students and teachers see the same five lines. This is a real profile shape:
The insight, not the number. This class communicates clearly and takes AI output at face value. The dashboard says exactly that, and suggests the lesson that addresses it.
One actionable next step per student, addressed to the student: “Ask where the 40% figure came from before you use it.” No praise, no scolding.