Microsoft 365
You're already paying for it. This is the day it starts paying you back.
Most companies I speak to have got Copilot. Ask them what problem they're solving with it and they go quiet.
The licences went out, an email announced it, and a handful of people had a play. Six months later finance is renewing a seat count that three people use properly.
I sat down with a financial services firm on a first meeting and counted 60 Copilot licences nobody was using. Their CRM and their accounting system were in much the same state.
The tool works. What's missing is anyone having sat down with your team and worked out which parts of their actual week it should be doing.
That's the day.
The Microsoft estate is where I see AI projects stall most often, and the reason is consistent.
Copilot arrives as a licence rather than as an answer to something. Nobody defined the problem it was bought to solve, so nobody can tell whether it's working. The people who might have used it opened it twice, got a mediocre answer because they typed one line into it, and went back to doing the job the old way.
That's a training and process gap wearing a technology costume.
Across 13 dated engagements, a client has weighed buying a new platform against configuring what they already own. Zero reversals on the record. The tool you've bought is almost always further from its ceiling than anyone in the room thinks.
So the day starts with the work. Your team brings the tasks that eat their week, we find the ones Copilot is good at, and we build the prompts for those jobs in the room. The prompts stay behind in a form the rest of the business can use.
One team, up to 10 people, the ones who'll use it daily. Remote or on site.
It works best when everyone in the room does broadly similar work, because the prompts we build together are then useful to all of them on Monday.
Right for you if
Not right yet if
Before the day, everyone writes down the three tasks that take the most of their week. Handwritten, not a template.
That single exercise does more than any module. It tells us where Copilot fits, and occasionally it tells us the honest answer is that it doesn't, in which case you've saved the cost of finding out slowly.
An honest map of Copilot's strengths across Word, Excel, PowerPoint, Outlook and Teams, and the places it produces confident nonsense.
Knowing the second half is what stops someone sending a Copilot summary to a client without reading it. Your team leaves able to tell the difference between an output worth trusting and an output worth checking.
Context is the whole game. The prompt is the delivery mechanism.
We work through the tasks the team brought in the morning and build real prompts against them: structure, context, refinement, and how to iterate when the first answer is thin. Nobody leaves with a generic prompt list off the internet.
The day produces a written prompt library specific to your business, and two or three people who understand it well enough to teach the next person.
Governance sits here too: what your team can put into Copilot, what stays out, and where your tenancy boundary sits. Plain rules, on one page. Where you need the layer behind that, it lives on Govern.
Written against your team's real tasks.
Where Copilot earns its licence in your business, and where it doesn't.
One page. What goes in, what stays out.
Named people who can carry it to the rest of the business.
Skills that stay when we go.
From £2,500. One day, one team, up to 10 people. Remote or on site. Headcount and scope drive the final number.
The day is built from the tasks your team brings in, so it calibrates to the room. Where a team is already competent, the day shifts toward the prompt library and the governance rules, which is usually the gap anyway.
The tool has to fit the environment. Where a business runs on Microsoft 365, Copilot sits inside the tenancy the data already lives in, and that counts for a lot. Where the work is heavy on writing, analysis and long context, Claude does more. We'll tell you which applies to you, including when the answer costs us the bigger engagement.
Covered on the day, in plain rules your team can follow without reading a policy. Where you need the full governance layer, an AI policy and the enforcement behind it, that's Govern and it's a separate piece of work.
One team at a time works better than one big session. Where you're rolling out across departments, the day is designed to leave champions behind who carry it, and we can scope a multi-team programme from there.
Then you'll hear that, and you'll hear it before you spend more on seats. I'd rather tell you on day one than take the training fee twice.