The model-makers themselves teach this first: tell AI WHO to be before you tell it what to do. "Answer as a lawyer" and "answer as a friend" produce two completely different answers to the same question.
The same question, answered by the right expert instead of a generic assistant — sharper and more useful every time.
This is the single most-taught prompting technique across Anthropic, OpenAI, and Google's own guides — because a role narrows what the model pays attention to, the same way telling a person "think like an editor" changes what they notice in your writing.
Open the AI you use (in any chat). This works in all the popular ones.
Copy the prompt below, paste it in, and replace anything in [brackets] with your details.
Read what comes back and ask it to adjust — shorter, warmer, more detail. Two tweaks and it’s perfect.
This is a one-off you can run in any chat. The real magic is making it repeat — ask AI to do it every week, keep it as a living system, or have it build you a tool that does it for you. That jump from "neat trick" to "runs my work" is exactly what the ∞ apps are for.
Read the full story → 5 prompting habits straight from AI's own makers (not internet folklore)
Plus the printable AI Moves Starter Pack (PDF). No spam, just things that make you go "I had no idea."