From individual use to team practice
Policies and support that actually hold.
There's a common pattern in small and medium teams using AI. One or two people pick it up. They get good at it. They start producing visibly better, faster work.
The rest of the team notices. Nothing else happens.
Six months later, the same two people are still the AI users. Everyone else is still doing the work the old way. The gap between them has grown. Nobody talks about it openly.
Closing that gap is doable. It means moving AI from a few enthusiasts to a shared way of working, without forcing anyone.
Why most teams get stuck
The failure is rarely about the tools. It's two silences facing each other. The people using AI don't share what they're doing, because it's working, they're busy, and teaching isn't their job. The people not using it don't ask, because asking exposes that they don't know where to start, and they already feel behind.
Everything usually thrown at that gap (the announcement, the training day, the policy) addresses neither silence. None of those moves are wrong. None of them are sufficient either, because none of them change what anybody does on Tuesday.
What doesn't move a team
Announcing that "we're going to start using AI"
A generic training day everyone enjoys and forgets
A policy written before anyone is using AI properly
Everyone on their own tool, on their own account
One person becoming the team's AI expert
What does
A shared paid tool, given to everyone by default
One recurring workflow redesigned with AI in it
Early users given time to show their work
Good examples made visible where people already work
Letting the resistant opt in at their own pace
The companies that get this right aren't the ones with the most ambitious AI strategy.
The stage most teams stall at
A rough model. First one or two people experiment privately. Then their use becomes visible, and others notice without engaging. Then the early users start showing colleagues, and a few more try it. Then the team has a shared paid tool, and some workflows involve AI by default. Finally it's embedded: shared prompts, shared standards, a capability rather than a curiosity.
Most teams stall between the second and third stage, which is precisely where the two silences hold. Getting past it isn't a matter of enthusiasm; it needs someone to make sharing part of the job. You'll know it worked when new hires are shown the AI workflows during onboarding, and when there's a set of shared prompts people actually reuse.
Standard, not mandatory
Every team has people who don't want to use AI. Some have good reasons, some are simply slower to move. Forcing them produces resentment and bad use, not adoption. The workable expectation has three parts: the tools are available, the workflows that use them are standardised, and anyone who wants in has everything they need.
There is one exception. Once AI use is the standard for a specific task, someone using a different process needs a reason. That isn't forcing AI on anybody. That's having a normal standard for the work.
Three things to do this week
01
Rebuild one real workflow, not the whole team
Pick something the team does weekly and redesign it with AI in it. Ambition spreads from evidence, not announcements.
02
Make the good examples visible
A channel where people paste what actually worked does more than a training day ever will.
03
Avoid the single AI expert
If everything routes through one person, they burn out and the capability leaves when they do.
Conclusion
01
Individual use doesn't spread into team practice on its own. It takes deliberate effort.
02
Get on a shared paid tool, rebuild specific workflows, and make good practice visible.
03
Don't force the resistant. This is a patient year, not an ambitious quarter.
Go to next