Signing in

You will be sent to MillerKnoll sign-in.

AI

Start with a Question, Not a Tool

The most common mistake in AI pilots is starting with the tool: "Let us try Copilot on everything." Useful experiments start with a question: "Will Copilot cut prep time for weekly pipeline reviews if I paste messy notes?"

Lesson 1

Curiosity with a finish line.

A good experiment question is specific, measurable in one sitting, and tied to work you already do. If you cannot describe what success looks like, you are not ready to prompt yet.

Questions beat tools because they force you to notice what you are trying to change — time, quality, consistency, or confidence — before you judge an output.

Core principles

  1. Bad question: "Can AI help sales?" Too broad; any output looks like progress.
  2. Better question: "Can Copilot turn my bullet notes into a table my team can scan in under two minutes?"
  3. Success metric examples: minutes saved, edits needed before send, errors caught, colleague usability rating.
  4. If the answer is no after a fair test, that is a useful result — you learned where AI does not earn a place yet.
  5. Write the question before you open Copilot. The discipline matters more than the wording.

Check yourself

What should you define before opening Copilot for an experiment?

Do this in Copilot

Write one experiment question for a task you did this week. Include what you would count as success.

Paste into Copilot Chat. Answer the follow-ups, then save your refined question.

Sharpen your experiment question

I am considering using Copilot for [TASK]. My rough question is: [YOUR QUESTION]. Ask me up to three short follow-up questions to make this question specific and testable in under 30 minutes. Do not suggest prompts yet.
Open Copilot →
  • Clarifying questions

Did you run this in Copilot? Mark complete when you have tried it.

Next lesson: Design a Test You Can Finish This Week →