Enablement · 02
Judge the file, not the number of prompts
Microsoft's Office in Copilot example is ordinary on purpose: draft a launch brief, model a budget, build a deck, update a team file. Word, Excel and PowerPoint inside Copilot return a file you can still open in the apps. A company new to AI should judge that file the way it judges any other draft.
What to compare
Compare the completed file with the team's normal editing and approval, not with a count of prompts. A short prompt that produces a workbook nobody will stand behind is a failed pilot. A longer prompt that produces a brief a manager marks up is a pilot you can learn from. Usage charts do not tell you which of those you had.
Keep the human edit
The point of the integration, as Microsoft described it on 25 September 2026, is that the output stays an editable team deliverable, not text to paste into something else. The habit to teach is: generate, open in the app, change the parts that are wrong, then send it through the path you already trust.
One fictional pack, no score
FrontierOrg, a fiction, refuses to publish a cycle-time saving it has not measured. Your first Office in Copilot trial should be just as dull. One brief. One owner. A note of what the person still had to fix. That note is the training material.
Monday
Produce one real file with Office in Copilot, then list what a colleague changed before it was fit to share. Bring the list to the next team meeting. Leave the prompt out of the story.
What would make you reject the file even if the prompt had been excellent?
Sources: Microsoft blog, 25 September 2026, Introducing the new Copilot (opens in a new tab).
Where this note mentions FrontierOrg, that is a fictional method for rehearsing a decision. It is not a company, not a client, and not a case study.