Pull vs Push Prompting

The two modes

• Push prompting is the way most people work. You do 80% of the thinking, write out the steps, and let AI finish the last 20%. You are telling it how to do the task.
• Pull prompting flips it. You give the outcome you want, then let AI figure out the path and pull the details out of you by asking questions.

The way to picture it: push is a friend giving you directions ("left at the church, red mailbox"). Pull is a GPS. You give the destination and it works out the route.

How to pull-prompt, step by step

1. 1
   Give it the role and context like normal. Who should it act like, and what is the situation. This is the same setup as the 4-part prompt formula.
2. 2
   State the outcome, not the steps. "I need an email sequence that turns cold leads into booked calls." That is the destination.
3. 3
   Tell it to interview you. Add: "Ask me every question you need to build this, one at a time, then give it back as a table or a doc."
4. 4
   Answer the questions. Tip: use the voice-to-text mic so you can talk through seven questions instead of typing them.
5. 5
   Refine by pulling more. "Ask me three more questions to make this sharper." Keep letting it pull until the output is right.

Why it is better

You stop guessing what to include. The tool surfaces the gaps you would have missed, and the answer is built from your real details instead of its generic defaults. This is the same idea software teams call spec-driven development: define the outcome, let the build follow.

A copy-paste opener

Act like a [expert role] who specializes in [your situation].
I need this outcome: [the result you want, not the steps].
Before you write anything, ask me every question you need to do this well,
one at a time. When you have enough, give it back as [format].