People stay in charge
The people who get real value from AI are not the ones with clever prompts. They are the ones with a steady method: define the task, give good context safely, work in stages, and test the result. This module gives you that method, so the work stays yours and the tool stays a partner.
By the end, you will be able to
- Define a task clearly enough that AI can help with it.
- Give useful context without exposing anything you should not.
- Work in stages instead of expecting one perfect answer.
- Test a result, and know when to stop and hand it to a person.
Lesson 1
Define the task
Most weak AI results come from weak requests. A clear task is most of the work.
Before you ask for anything, get clear on what you actually want. A vague request gets a vague answer. A defined one gets something you can use.
- Purpose: what is this for?
- Audience: who will read or use it?
- Outcome: what does a good result look like?
- Constraints: length, tone, what to avoid, what to include.
- Authority: who has the final say on this?
Weak: write something about our program. Clear: draft a short, warm paragraph for a first-time donor newsletter, describing our after-school program in plain language, under 90 words, no statistics unless I provide them.
Try it
Take a request you might make and add the five pieces: purpose, audience, outcome, constraints, authority. Notice how much better the ask becomes.
RememberA clear task beats a clever prompt. Name the purpose, audience, outcome, constraints, and who decides.
Lesson 2
Give useful context, safely
AI does better with good context. The skill is providing it without crossing the safe-use lines from Module 3.
Good context is the difference between a generic answer and a useful one. Examples of what you want, source material to work from, and clear definitions all help. So does saying plainly what to leave out.
Note
This is where Module 3 comes back. Provide the context the task needs, and no more. Strip the private details, use approved tools, and prefer aggregate over identifying information.
Try it
For a task you have in mind, what is the smallest set of context that would make the answer good, without including anything private?
RememberGive the context the task needs and nothing private. Good, safe context beats a big, risky dump.
Lesson 3
Work in stages
Do not expect one perfect answer. Treat it as a back-and-forth you steer.
The strongest way to work is in stages, keeping your hand on the wheel the whole time. Each stage is a place where your judgment shapes the result.
- Plan: agree on the approach before generating.
- Generate: get a first version.
- Critique: ask what is weak, missing, or unsupported.
- Compare: get an alternative and weigh them.
- Revise: improve with your direction.
- Document: note what you used, checked, and decided.
Try it
Next time, resist accepting the first answer. Ask it to critique its own draft, then revise. See how much the second pass improves.
RememberWork in stages you steer: plan, generate, critique, compare, revise, document.
Lesson 4
Test the result, and know when to stop
A good result earns its place by passing a test. And some tasks should be handed to a person, not finished by AI.
Before a result becomes real work, test it against something outside the tool: a rubric, an independent source, a counterexample, or a person with real expertise. If it cannot pass that test, it is not ready.
Try it
Name one kind of task in your work where the right answer is to stop and bring in a person. Naming it now makes it easier to do later.
RememberTest every result against something outside the tool, and stop when stakes, authority, or sensitivity call for a person.
What you leave with
Reusable task brief and verification checklist
A short, reusable brief that captures the five parts of a good task and the checks a result must pass. It makes the human-led method repeatable across your whole team.
This is just for you. It saves on this device only, and nothing is scored.