What People Actually Need to Be Able to Do
Why this matters
Most enablement starts with tools or content: here is the product, here are the prompts, here is the library. Better enablement starts with the decisions people must make — because the point of the whole program is that people decide better, not that they attended more sessions or collected more prompts. If the decisions are not named first, the training is solving a problem nobody confirmed exists.
What you will be able to decide after this
- What the group actually faces: the decisions, not the tools.
- What good judgment looks like on those decisions — specifically enough to be recognized.
- What evidence would show the capability exists, and what happens if it does not.
Core lesson
Ask four questions, in order:
- What decisions will these people face that involve AI? The real decisions in the real work — the weekly update, the ticket triage, the draft that goes to a customer, the recommendation that moves money. Not "AI decisions" in the abstract; the decisions that already happen, with AI added.
- What does good judgment look like in those decisions? Written so a manager and a peer could both recognize it: the constraint set before the prompt, the verification performed before the output is used, the stop called when the net value turns negative.
- What evidence would show they can exercise it? The artifacts and behaviors that demonstrate the judgment — a verified output, a constraint written for a new task, a decision to stop, a post-mortem that names what went wrong.
- What happens if they cannot? The cost of the capability not landing. This is what makes the statement matter to the people who fund it: if the judgment does not transfer, the errors continue with the tool in the middle of them.
Training that does not change decision quality is just activity. The capability statement is the antidote: it names the decisions, the standard, and the evidence — before any session is designed.
Worked example
An operations team of twelve is about to get AI access for drafting customer replies. The capability statement, one page: the decisions — which drafts to use as-is, which to revise, which to refuse and route to a person, and when to stop drafting with AI entirely. Good judgment looks like: constraints set before drafting (tone, length, facts allowed), verification run on every draft that goes out, the ten-question filter applied to anything new, and a clear call when a customer case is too sensitive for an AI draft. Evidence: a two-week sample of drafts showing the constraint and verification pattern, a new-case exercise where the person applies the filters unprompted, and a post-incident note that names the judgment call and the lesson. What happens if they cannot: the 18% rework rate from the business case does not move, and the first sensitive-customer case goes out wrong with a tool in the middle of it. The statement gets reviewed with the team lead, who catches the missing decision — "drafts for high-value customers always wait for a person, no exceptions" — and it is added before any training is designed.
Practice
For the group you will enable, write the four questions and your first answers: decisions, good judgment, evidence, cost of failure. Mark the questions you cannot answer yet — those are what the review is for.
Apply — produce the artifact
For one real group you need to enable, write a short capability statement: the key decisions they must handle, what good judgment looks like on those decisions, and how you would recognize it in their work.
Verify
Review the statement with someone who currently does the work or manages it. If they say the decisions or the standard of judgment are unrealistic or incomplete, revise.
Sources
This module is original practice guidance based on the authoring standard and does not depend on a specific external factual claim. Editorial review is still required.