Copilot did not do my job this fortnight. It did save me from another unnecessary argument with a blank page.
And several rounds of “I know what I mean, why does this sound so complicated?”
There was no single dramatic moment where AI swept in and transformed the working week. The value was in the smaller moments: getting an idea into usable words, working through a solution, finding my way past the bit that was holding everything up.
That is what I want this regular feature to show. What I worked on, what helped and where I still had to do the thinking.
Consider this the fortnightly view from behind the laptop.
Getting the words out of my head
I used Copilot for emails, communications, presentation wording and meeting introductions. I also used it to work through training explanations and prompts for other AI tools.
The common problem was not that I had nothing to say. It was getting from the version in my head to a version someone else could use.
Sometimes I needed a starting point. Sometimes the draft existed but sounded as though it had swallowed a technical manual. Sometimes it was simply too long.
“Make it shorter” did a fair amount of work.
The useful part was having something to react to. It is much easier to say “That is not what I mean” when there is a sentence in front of you than when the entire document is still a blinking cursor.
Making the explanation earn its place
Knowing a topic well can make explaining it surprisingly difficult. You can see all the detail. Your audience needs somewhere to start.
I used Copilot to think through the audience and learning outcome before structuring presentations and training. What should someone understand by the end? What needs explaining first? Which detail is helpful, and which detail am I including because I happen to know it?
Two of my actual instructions were:
Explain this in layman’s terms.
And then:
Make it even simpler.
There is a useful professional development exercise in that second prompt. If the explanation still needs translating, it probably needs another pass.
I also worked on slide structure and presenter wording. The aim was a story a person could follow, rather than a collection of things I could technically tell them.
Talking through the build before building everything
I worked through ideas involving SharePoint lists, pages, metadata, dashboards, forms and automation.
That is planning work. I am not presenting a list of six newly deployed solutions with matching ribbon-cutting ceremonies.
The most useful way into the conversation was often to describe the experience I wanted:
When someone does this, I want this to happen next.
From there, I could explore what components might be needed and which approach made sense. Starting with the person using the solution helped me avoid getting attached to a particular component before I had settled the problem.
I kept coming back to questions such as: what should happen when someone clicks this? Where should they go next? What information do they actually need to see?
“How do I make this simpler?” was a recurring theme. It usually is.
Getting unstuck halfway through
Not every useful AI conversation starts at the beginning of a project.
Sometimes it starts here:
This part works, but this part doesn’t. How do I fix just that?
That was another use this fortnight: working through a specific problem without treating the whole thing as a fresh build.
I wanted help isolating the troublesome part, understanding possible causes and considering a next step. A proposed fix still needed checking against what I was actually working on.
The value was having a troubleshooting partner available at the point where I would otherwise have gone around the same loop again.
Finding the thing we had already done
There was also the familiar workplace mystery: I know we did something on this before, but where is it?
I used Copilot to help rediscover earlier work and bring relevant information back into view. What you can retrieve depends on the Copilot experience, connected work context, licensing and access available to you. It is not a promise that every chat box can search your entire organisation.
For me, the useful outcome was getting back to the work rather than spending the conversation trying to remember where the work had gone.
I also used AI to consider themes, important numbers and what a management audience might need to focus on. Suggestions helped me organise the material. I still had to decide whether the interpretation made sense.
The follow-up is part of the work
My conversations included “I don’t like that”, “That’s too technical”, “Give me another way” and “What am I missing?”
I do not see those as evidence of a failed prompt. They are how I refine the result.
I know the audience, the context and what I am trying to achieve. If an answer misses that, accepting it because it arrived neatly formatted would be a strange decision.
So what did Copilot actually save me from?
Mostly, the friction around my job. The blank page. Another rewrite. A search for something we had already created. Getting stuck on the same part of a solution.
The work still required judgement. It just did not need quite so much wrestling before I could get on with it.
Next time you catch yourself thinking “There has to be an easier way to do this”, try opening your approved AI tool and describing that exact problem. Start there. You do not need a spectacular use case to have a useful one.
