It’s Monday morning and you’ve just announced an AI plan. People nod and ask polite questions, then go back to their desks and wait for proof of what this means for their Tuesday. I think leading AI adoption comes down to that proof. It arrives in the first two weeks, and it comes through your hands far more than your memo.
Canadian firms are learning this on the job. Statistics Canada found 32.0% of businesses using AI trained existing employees on it in the year to spring 2026. For everyone else, the first two weeks of the rollout are the training. Some of the people in that room have used these tools already. In Alberta, 21% of workers used generative AI in their main job between September 2024 and July 2025, against 22% nationally.
Nobody on your team is grading the announcement. They heard the words, and words are cheap in a rollout. What they are watching is the small evidence that lands next: which task changes first, whose work gets easier, whose pile grows by an hour of checking, and if the person who announced the plan is anywhere near the tool once the meeting ends. By week two, every one of those questions has an answer.
What does your team actually judge in the first two weeks?
They judge what lands on their desk. The first workflow the plan touches tells them more than the launch meeting did. If the first job the tool takes is one everybody hated, run it on real work the team already owes, and when it comes back done properly, the plan earns credit the memo never could.
If the first job arrives half wrong and lands on a person to fix by hand, the plan spends that credit on the spot.
They also judge the gaps. A tool that saves one person an hour and hands another person an hour of checking has not saved the firm anything, the saved hours that quietly disappear, and the second person knows it before you do. Staff do that math at their own desks, with their own workload as the ledger.
Your team forgets the speech by Friday. They remember the first task the plan put on their desk.
Why does your own use matter more than your announcement?
Because your team takes its cue from what you do where they can see it. If you announce AI on Monday and spend the week working the old way, the lesson lands without you teaching it. The plan is a poster.
If you run your own first drafts, your own summaries, your own weekly numbers through the tool, and you let people see the rough edges as you fix them, the lesson is the opposite one, and just as free. I’d start with your own weekly numbers.
Going first does something else as well. You learn where the tool stumbles on your firm’s real work before you ask anyone else to trust it with theirs. The complaint that reaches you in week two is one you have already met at your own desk, so you can answer it from experience instead of defending a decision you made in a meeting room.
Your team believes what it watches you do on a Tuesday, not what you said on the Monday.
What should the first workflow be?
Pick one the whole team can watch landing. Small enough to run inside the two weeks, visible enough that nobody has to take your word for the result, and close enough to the daily work that people can picture their own version of it.
A report that used to eat a morning is a good shape. A client letter nobody liked writing is a good shape. A trophy project six months out is not, because by the time it lands, the verdict is already in.
Run it in the open, first attempt, correction, and finished piece side by side. That viewing is the training, if you call it that or not. It is the same order I argue for with people. Judgment first, tools taught on real work.
The first workflow is the one the team can watch working from the next desk over.
What do you name as staying the same?
Name it plainly, and name it in the first two weeks while the judging is live. The finished-work standard does not drop because a tool drafted it. The person who signs a piece still owns it. The clients who trusted a name still get that name.
And the jobs your people do well are not on a table somewhere being priced. Say the true parts out loud, in words a person can repeat in the lunchroom.
A named list settles more than a promise does, because a list can be checked. It also shapes the team you are building. The juniors watching you run this are learning how change gets led here, and that lesson outlasts the tool. I made the longer version of that point when I asked who becomes your senior if the early work all goes to the machine. The seniors you want in five years are forming their habits in weeks like these.
A written list of what stays the same is worth more than a promise that nothing will change.
My take: the owner’s own Tuesday is the real rollout.
What moves start your AI rollout this week?
Three moves, all inside the two weeks. Choose the first workflow and run it where the team can watch it work. Use the tool on your own work first, in the open, rough edges showing. Then write the no-change list, read it to the team, and post it where the work happens.
None of that needs a budget meeting. It needs you to go first and be seen going first. If you want those two weeks planned with the same care as the tool choice, my AI training workshops start at the leader’s own desk and work outward from there.
Week two is the real announcement. The memo was only the rehearsal.
Leading AI adoption, quick answers
How do you lead AI adoption on a small team?
Go first, in the open. Run your own drafts and summaries through the tool where people can see the rough edges, pick one first workflow the whole team can watch land, and name out loud what stays the same. Your team judges the plan by week two, not the announcement.
What should the first AI workflow be?
One small enough to finish inside two weeks, visible enough that nobody has to take your word for the result, and close to daily work. A report that used to eat a morning or a client letter nobody liked writing works well. A trophy project six months out does not.
How do you ease staff worry about AI?
Write down what stays the same and read it to the team. The finished-work standard doesn’t drop, the person who signs a piece still owns it, and clients still get the names they trust. A list people can check settles more than a promise that nothing will change.
Read your no-change list to the team this week. The AI Newsletter with ZAK covers rollouts like that every week.
4 Comments