One brief, three routes

This is the whole course, compressed. Take the Lesson 1 brief. Route A: you, no Copilot, 20 minutes. Route B: Copilot Chat. Route C: Researcher or Analyst if you can open them; if not, Chat with “list sources I should open.” Compare. Send one. Throw the other two away.

Then pick, verify, send.

You are not looking for the fanciest output. You are looking for the one you can stand behind in the room.

Core principles

  1. Hold the brief constant. Do not improve the prompt between routes except to name the surface.
  2. Score: completeness, traceable facts, voice you would actually send, time to a defensible draft.
  3. If Route C is missing, two routes still count. Write that on the card.
  4. The card you keep: “For [this job], start in [surface], because [quality]. Verify [what].”

Related: The Taste Test — fair compare inside Chat

Check yourself

What should you keep after a three-route test?

Do this in Copilot

Send the brief (or schedule the send). Pin one routing sentence next to your Taste Test card if you have one.

Paste outputs labeled A, B, C (or A, B). Do not say which surface produced which until after it scores.

Score the three routes

I ran the same brief three ways (or two if an agent was missing). Brief: [ONE SENTENCE]. Score A, B, and C 1–5 on: completeness, traceable facts, sendable voice, time-to-useful. Do not guess which Copilot surface wrote which. Rank them. Then I will tell you the mapping. After I reveal it, write one routing sentence: For [job], start in [surface], because [quality]. Verify [what].

Techniques: Blind scoring

Open Copilot

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

AI

Part of the MillerKnoll collective

Studio · 2026