I've worked with about a dozen organizations on AI adoption in the past year, and the pattern is so consistent it's almost boring to describe: the technology works fine. The conversation about the technology is where everything falls apart.
5 mistakes come up in nearly every company, and they're all communication failures, not technical ones.
They announce the tools before they address the fear. One org sent a company-wide email about their new AI initiative on a Monday. By Wednesday, three people had asked their managers if their roles were being eliminated. If your team hears "we're rolling out AI" before they hear "your job is safe, and this is about freeing up your time" — you've already lost them. The fix is pretty simple: lead with the human message. Say it first, say it clearly, and say it before you mention the tool.
They train on the technology instead of the workflow. I keep seeing the same generic "Intro to AI" workshop where everyone sits through a prompt engineering deck. And then... nothing changes. Because nobody showed the person who reconciles expenses how AI helps with expenses. Nobody showed the person who writes donor reports how AI helps with donor reports. The training has to start from the work. Take five real tasks from five real roles on the team, build the training around those, and suddenly people get it.
They let the gap between early adopters and everyone else become a canyon. Every org has that person chatting with their meeting transcripts at 11pm. (You know who they are.) That enthusiasm is valuable, but if it's not channeled into structured peer teaching, it becomes intimidating. The fix: give your early adopters a formal role. Make them mentors with a specific scope — three people each, one workflow at a time.
They measure access instead of adoption. I asked one client how their rollout was going. "Great, we gave everyone a license." Cool. How many are logging in? "...we don't track that." (sigh) 30 licenses and 30 people actually using AI in their daily work are completely different achievements. Track login frequency, track which workflows are changing, and track whether people are teaching each other. That last one is the real signal.
They start too ambitious. Task forces, strategy decks, 12-month roadmaps — and then the first deployment reveals that half the team has gaps with basic technology. One org I worked with started their AI journey with a project management tool. That single, boring rollout taught them more about their team's readiness than 6 months of planning ever could have.
The through-line across all five is the same: organizations treat AI adoption as a technology project when it's actually a people project. The rollout plan, the training design, the metrics you track, and the pace at which you move all need to account for the humans in the room.
If your adoption numbers are stuck, look at how you're talking about AI before you look at what you're deploying.
— Jamie
P.S. If you recognized your own organization in any of these, you're not alone and you're not behind. Most companies are making at least 3 of these mistakes simultaneously. Reply to this email and tell me which one hit hardest. I'm collecting examples for a talk I'm building on this topic.