Most weak AI results come from messy context. Long explanations and half‑formed ideas confuse the system. This tutorial shows a simple fix: use multiple‑choice questions to create clean context before the AI responds.
Define Goal
Write one line describing the final outcome.
Example Goal: “Create 3 strong taglines for my AI newsletter.”
Set Task
Tell the AI its job is to interview you to build context, not guess.
Example Task: “Interview me using multiple choice questions to collect the context you need.”
Control flow (Next Steps)
Give clear rules so it asks questions first, then waits.
Example Next Steps:
- Ask 8 to 12 MCQs
- Each has A, B, C, D options
- After the last question, stop and wait for my answers
- Do not generate outputs before I reply
Use interview prompt
Prompt example:
“Goal: Write a landing page headline + subheadline for my AI prompts course. Task: Interview me to build the minimum required context using multiple choice questions. Next Steps: Ask 10 MCQs with A to D options. Keep questions short. After Q10, stop and wait. Do not write anything until you get my choices.”
Reply with letters
Answer like: “A, D, B, B, C, A, D, C, B, A”. This removes miscommunication and keeps context tight.
Generate concepts from choices
After you answer, tell it to produce a small set of options based on your selections.
Example: “Using my choices, generate 4 concept options with a name and 2 line description each.”
Compress previews with grid
Once it generates concepts, ask for a compact comparison to save tokens.
Prompt example:
“Generate a 4x4 grid, one concept per grid cell, with name, 1 line idea, and 3 keywords. Then ask which cells I want as standalone versions.”
Note: If the output still feels off, do not rewrite everything. Ask for 3 more MCQs targeting what is missing.