Instead of pretending you gave it enough context, you have the AI interview you until the task is clear.
Think about a contractor remodeling a room. A bad one hears “make it nicer” and starts swinging a hammer. A good one asks first: What’s the budget? Who uses the room? What does done look like? Interview prompting makes the model act like the good contractor before it touches the work. A strong prompt can admit you don’t know everything yet. The line we use: ask me any questions one at a time, to gain as much context as you need to succeed at this task.
How it shows up
This is most useful when the work is fuzzy: a proposal, a workflow, a client email. The missing context is often in your head. You might start with, “Draft a follow-up email after my strategy call. Ask me questions one at a time until you have enough to write it well.” Now the AI isn’t guessing the tone or audience; it asks, you answer, and the context gets built through the conversation. It’s part of a larger prompt framework habit, CRIT (context, role, interview, task), and the “one at a time” part matters, because twelve questions at once means you skim, answer badly, or quit.
Why you care
Most bad AI output comes from the model missing the real constraints, not from the model being useless: you forgot who the audience is, or that the client is sensitive about pricing. Interview prompting lets the AI slow down before it speeds up, and keeps you in the human in the loop where your judgment matters most: defining the work. Start the conversation like a person and let the AI ask for what it needs.