🧭 Plot Summary
There's no new content here. This problem puts the unit together. Your team meets a client and learns what their work looks like: the tasks, how often, what a mistake costs, and who checks today. You'll sort every task with the four questions, decide where the tool should do the work, where it should help, and where a person must stay in charge, and say why.
Then you plan the checking and the failures: who looks at the tool's work, when and how, and what happens when it's wrong in a way that looks right, or when it stops. You'll write an AI-use guide and a log the client can keep, have another team test them, revise, and defend your choices out loud.
Resource Map: where each piece comes from
Stuck on one step? Jump straight to the lesson that covers it.
What a tool collects, who else sees it, and what should never go in
The four questions, the four levels, and the two stop signs
The four strands, and the difference between a claim and evidence
The camera test for tasks, questions that get true answers, and safe interviews
Can versus should, naming your evidence, and saying what you don't know
One concrete ask, supported by evidence and fair to the other side
✅ Self-Check Before You Roll On
Check off each item as you get there. These aren't grades. They're your own signal.