The AI Assistant Nobody Knew They Had
There’s an AI assistant sitting inside your organisation’s Microsoft 365 licence right now, and a decent share of your people have no idea it exists. If you’ve been asked to look into Microsoft Copilot training, that fact is the best place to start, because it changes what kind of problem this is.
It also explains a lot about the Copilot conversation happening around you: everyone in it is right. Leadership is right that these tools matter, and that an organisation which learns to use them well will feel the difference. Your colleagues who tried Copilot once and drifted away are right too, because the answer they got genuinely wasn’t very good. And if you’re the person in HR or L&D juggling all of this alongside a full plate of very human work, you’re right to want a solution that doesn’t become yet another programme to run.
The good news is that all three of those things can be true at once, because the missing piece is small, specific and teachable.
Why has adoption stalled?
Because access and adoption are two different things, and most rollouts only deliver the first one. According to Recon Analytics research covering more than 150,000 US workers, fewer than 4 in 10 employees with Copilot access actively use it. That number gets read as resistance, but look closer and it breaks down into three very ordinary gaps.
The first is awareness. Copilot Chat comes included with Microsoft 365 work accounts, which means many of your people are sitting next to an AI assistant they don’t know they have. The second is usefulness. Even those who’ve found it often can’t picture what it’s actually for, beyond a vague sense of “AI stuff”. And the third, easy to miss, is permission. Plenty of people who are genuinely curious hold back because nobody has ever said out loud that they’re allowed to use it for real work.
None of that is a workforce saying no to AI. It’s a workforce that was handed a new colleague and never introduced properly.
Why do first attempts with Copilot disappoint?
Most people prompt Copilot the way they’d type a search: a short request like “write me a report”. Copilot doesn’t know who the report is for, what it needs to achieve or what good looks like in your organisation, so it guesses. The output is generic, and a busy person quite reasonably concludes their time is better spent elsewhere.
Nobody did anything wrong here. They were simply never shown how to get the most out of the tool.
What’s the fastest way to get better results from Copilot?
Here’s that shift, and it’s yours to share with your teams today. Before the next Copilot request, add this line:
“Before you answer, what questions do you need to ask me?”
We call it the Not Yet Method. Instead of guessing context, Copilot asks about the audience, the goal, the constraints. Answer those, and the output transforms on the very next try. For someone whose first attempt fell flat, that one moment reopens the whole conversation.
It costs nothing, takes a minute to share, and gives you an immediate, visible win to build on.
What does good Microsoft Copilot training look like?
Short, practical and light to organise, because your diary is already full and this shouldn’t become a project.
Our Smarter Work with Copilot workshop takes two hours and can be delivered whether you have standard or Premium access. People spend it hands-on with their own real work, their meetings, their emails, their reports, rather than watching demos of someone else’s. They learn four prompting techniques (the Not Yet Method is one), find the everyday tasks where Copilot saves them the most time, and leave with a plan for their first three uses that month.
It also closes all three gaps at once. People discover the assistant already sitting in their licence, see exactly what it’s useful for in their own role, and hear something surprisingly powerful: an official, organisation-backed yes. When training is put on for a team, the permission question answers itself. This tool is yours, this is what it’s for, and you’re encouraged to use it.
It also honours the instincts of your more cautious people, which are good instincts. The session is clear about what Copilot cannot do: make decisions, replace expertise, take responsibility. You are the pilot, Copilot is the co-pilot, and human judgement stays at the centre throughout. Careful adopters tend to become the best adopters once they can see exactly where the boundaries sit.
The evidence says this kind of structured session is exactly the gap to fill. McKinsey found that only 1 in 3 organisations offer structured generative AI training, despite nearly half acknowledging it’s essential for adoption to succeed. Your organisation has already done the hard part by putting the tool in everyone’s hands. Training is the short bridge between owning it and using it.
