Wishlists are easy and mostly unused. “I could use Copilot for email, decks, and analysis” is true for everyone and useful for no one.
Discovery is an interview, not a sticky-note wall.
This course treats Copilot as an interviewer. It asks where your time goes, what you repeat or dread, and what you already do well — then it is allowed to drop ideas that are generic or impossible with your access.
Your job is honesty. Copilot’s job is to refuse a boil-the-ocean backlog.
Core principles
- Near-term experiments beat multi-month transformations.
- Already-working uses get demoted. Do not rediscover what you already do.
- Hard limits count: no admin rights, no certain data, Copilot only — say so up front.
- If the output could be given to any MillerKnoll title, the context is too thin.
If the list would be generic: Your AI Identity
Check yourself
Why skip a brainstormed AI wishlist?
Wishlists feel productive and then sit. An interview with hard filters produces a shortlist you can run.
Do this in Copilot
Write three facts Copilot must know before it ranks anything: your role, your tools, one thing you already do well with Copilot.
If Copilot says it knows almost nothing, open Your AI Identity before Lesson 2.
Check context thickness
I am about to run a Copilot use-case discovery interview. Based only on what you already know about me in this chat or in standing instructions, say: (1) what you know that is specific, (2) what is still generic, (3) two grounding questions you must ask before ranking anything. If you know almost nothing, say so. Do not suggest use cases yet.
Did you run this in Copilot? Mark complete when you have tried it.
RecordedNext lesson: Run the Discovery Interview →
Navigate: press j for next lesson,k for previous.