Test It in a Fresh Chat

Open a brand-new Copilot chat. Do not paste the brief. Ask a question that only works if Copilot already knows you — something about this quarter’s priorities, your audience, or a constraint you wrote down.

If a new thread still sounds generic, it did not load.

If the answer could still fit anyone in your title, the identity did not load, got truncated, or is still too vague. Fix the setting or thicken the brief. Do not add more prompt text as a workaround yet.

A passing test is one line Copilot now gets right that it could not get right last week.

Core principles

  1. Fresh chat. No paste. That is the only honest test of standing context.
  2. Pick a test question that requires a load-bearing fact from your brief.
  3. If it fails: check Settings first, then shorten or sharpen the brief, then retest.
  4. Memory, instructions, and project context are different stores. If one is empty, try the field you actually used in Lesson 3.

Check yourself

How do you verify that standing identity is working?

Do this in Copilot

Run the test question in a new chat. Note one line Copilot now gets right — or the setting you still need to fix.

Open a new Copilot chat. Send this with no extra context.

Fresh-chat test

Based only on what you should already know about me from standing instructions, answer: what am I optimizing for this quarter, who is the primary audience for my drafts, and what should you stop assuming? If you do not know, say you do not know — do not invent a plausible professional.

Techniques: Claim spot-check

Open Copilot

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

Next lesson: Teach Copilot to Push Back →

AI

Part of the MillerKnoll collective

Studio · 2026