Where you're stuck
Why is your Copilot rollout stuck at proofreading emails?
Copilot adoption usually stalls here because people never get protected time to learn AI on their own work, and nobody has said what it's for. So busy people reach for the easiest use they can find, and email is always in front of them.
It's an easy stage to stay in, because the usage dashboard looks healthy. Organizations often burn months here, waiting for the first round of training to pay off, while the work itself stays the same.
What you've built
You bought the licenses and ran the training. People log in, and plenty of them like it. (Your team's emails have never been so well proofread.)
The thing most likely holding you back
People are learning AI on top of a full day job, with no time set aside to try it on their own work, so they pick whatever takes the least time. And there's no AI North Star, so everyone's effort points in a different direction.
What this stage costs
Melissa · 54 sec
Stop asking your team to learn AI on weekends
You keep paying for licenses while the work stays the same. A small group of enthusiasts runs ahead, and everyone else waits for someone to show them what AI is for.
An AI transformation lead at a mid-sized company told me Copilot was mandatory there, measured by tokens. A few engineers set up prompts that fire all day to climb the usage leaderboard. (Yes, there's a leaderboard.) The dashboards look great, and nobody trusts them.
Meanwhile, the people you're counting on to learn AI are already stretched.
80%
of workers say they lack enough time or energy to do their work. Source: Microsoft Work Trend Index, 2025 (31,000 knowledge workers, 31 markets, Feb to Mar 2025)
What I hear from leaders
Questions leaders are asking about Copilot adoption
We've trained thousands of people on Copilot. Why hasn't the work changed?
Where's the return on the licenses?
When are people supposed to find time to learn this?
Why the usual fixes stall
Melissa · 39 sec
The 'Piano Trap' of AI Adoption
The usual next step is more training, like prompt libraries and lunch-and-learns. Those teach the tool on sample tasks, away from anyone's real workflow. And AI takes practice, like a piano, and most people have no practice time left in their week.
More training also leaves the big question open: which work should AI change first? That call can be made for the whole company or for one team, and until someone makes it, people use better prompts on the same emails.
The other usual fix is the stick: put AI usage in performance reviews. Duolingo went AI-first and did exactly that in 2025. A year later, its CEO walked it back after employees wondered whether they were using AI for AI's sake. Source: Fortune, Apr 2026
What the research says about AI adoption at work
Only 42% of employees say they know how to identify where AI can improve their work. Employees are five times as likely to be top AI users when AI solves their work frictions. Source: Gartner, Oct 2025
Measured by impact, employees who get clear direction outperform employees who have more access to AI tools. Of frontline employees who use AI regularly, 66% get limited or no guidance on what to do with the time they save. Source: BCG, AI at Work, Jun 2026 (close to 12,000 frontline employees, managers and leaders)
48%
of the difference in which companies captured value from AI came down to organizational readiness. Personal readiness explained 25%. Source: McKinsey, Jul 2026 (750 employees, Feb to Apr 2026)
Start here
Give people protected time to learn AI on their own work
Everyone has Copilot, learning happens after hours, and each person decides alone what it's for.
A team spends a protected day learning AI on a workflow it owns. People leave able to read their work through an AI lens, pick the use case worth building, and rebuild a workflow together, pointed at a goal everyone shares.
From the field
One leadership team's shift from tools to a North Star
At a global education company, the AI conversation had mostly been about tools: lots of small pilots, a newly hired data and AI director, and a literacy survey where people asked for training, guardrails and a list of approved tools.
In the session, we asked about their North Star. The room went quiet at first. The questions shifted after that, from which tools to use to which business goal AI should serve.
One executive asked if solving an expensive problem counts as a North Star. (Yes!) The next day, the CEO asked the team to set one. A few weeks later, they were drafting candidates to sit above their strategic plays, something their OKRs had never had.
Two ways to start
Most teams at this stage need both. Start with whichever is missing most.
Make time to learn
The Applied AI Workshop
Six hours for one team and one real workflow, in the AI tools you already have. People learn AI on their own work, and they keep the skills that carry to the next workflow: reading work through an AI lens, screening for the use case worth building, and redesigning a workflow together.
Set the direction
The AI North Star Workshop
In two hours, up to ten people set an AI North Star for their part of the organization: the whole enterprise, one business area or a program team. They write the AI philosophy behind it, test a filter anyone can use on new AI ideas, and leave with the North Star in one sentence and a named owner to carry it forward.
Questions about Copilot adoption
How do you improve Copilot adoption?
Give people protected time to learn AI on their own work, and a direction for it. Learning sticks when people practice on a workflow they own, and a North Star tells them which work to start with. Gartner found employees are five times as likely to be top AI users when AI solves their own work frictions.
Which should we start with, the Applied AI Workshop or the AI North Star Workshop?
If people have never had time to try AI on their own work, start with the Applied AI Workshop. If teams are busy with AI but pulling in different directions, start with the North Star. Most teams at this stage need both, in either order.
What is an AI North Star?
It's one goal for AI that everyone in a group can check their own AI use against. A leadership team can set one for the company, and a function or team can set one for its own work. It answers the question most people have: what's AI for, here?
How is an AI North Star different from a vision or mission?
Your vision and mission say why the company exists and where it's headed. An AI North Star is narrower: the one goal your AI work serves, specific enough that anyone can hold an AI idea up against it.
Why do employees mostly use Copilot for email?
Email is in front of them every day, and drafting and summarizing are what the tool makes easy. When people are learning AI between meetings, they reach for the quickest use.
Why don't people make time to learn AI?
Most of them have none to spare. In Microsoft's 2025 Work Trend Index, 80% of workers said they lack enough time or energy to do their work. Learning AI takes practice, so the time has to come from somewhere, usually by taking something else off people's plates.