How do I get AI to ask me questions before it answers?

What it is

Generic AI output usually comes from missing context, not bad wording. Inviting questions makes the model tell you what it needs. Dan Martell calls this “reverse prompting”. It’s a practitioner technique, not a measured one.

How it works

1. Describe the task and goal briefly.
2. Add an instruction to ask clarifying questions first, for example: “Ask me any questions you need to fully understand this before you start.”
3. Answer the questions, then let it draft.
4. Save the questions it asks as a checklist for next time.

What the vendor guides say (checked 2026-10-17). Anthropic’s prompt-engineering guide stresses giving clear context, and tells agents to ask the user before hard-to-reverse actions. We didn’t find a vendor guide that explicitly recommends having the model ask clarifying questions before drafting. Treat this technique as a practitioner tip.

Benefits

• Better first drafts for client-specific work.
• Teaches you which context matters.
• Works in any assistant.

Limits and risks

• Adds a round trip, so skip it for simple tasks.
• Models can ask unnecessary questions. Tell it to limit itself to the most important ones.
• Answers you give become part of the prompt. Don’t paste sensitive data you wouldn’t otherwise share with that tool.