Free lesson · Skill · 5 min
Give it a role
Lesson 4 of 12 in AI at Work Foundations
The one idea
The model has absorbed the writing of lawyers, marketers, engineers, teachers, editors — everyone. A role tells it which of those voices to use:
"You are an experienced hiring manager reviewing a CV for a marketing role."
One sentence, and suddenly the feedback cares about the things a hiring manager cares about.
Why it works
Remember lesson 1: the model predicts text from patterns. "You are a [role]" narrows the pattern space to text written by that kind of person, for that kind of audience. It changes:
- What it pays attention to — a CFO role surfaces costs; a customer role surfaces confusion
- Tone — a coach encourages; an editor cuts
- Standards — "you are a skeptical reviewer" produces far more useful criticism than "any feedback?"
The pattern
You are [role with relevant experience].
[Context about the situation.]
[Task.]
Strong roles are specific: not "you are a writer" but "you are an editor at a plain-language consultancy who cuts jargon ruthlessly".
A trick worth stealing
Use roles to get multiple perspectives on the same thing:
"Review this proposal three times: once as a skeptical CFO, once as the customer, once as a lawyer. Give the top concern from each."
Three meetings worth of feedback, thirty seconds.
Your win today
Pick something you are working on. Ask your AI tool to review it in the role of the most demanding person who will actually see it. That review is your rehearsal.