Everyone treats AI adoption as a communications problem. It isn’t. It’s a change management problem — the same one we’ve always had — and we’remaking the same mistake we’ve always made.
One of my first jobs was as a change management specialist, and what I learned there has never stopped being true: when it comes to change, everyone wants to focus on communications. Communicationsis the easy part.
Take a migration a lot of people have lived through — moving a company from PeopleSoft to Workday. Those projects teach the same lesson every kind of change does. The hard part isn’t the announcement. It’s understanding the delta — the actual difference between today and tomorrow — and being able to articulate it accurately. That was the hard part then. It’s even harder now.
The four swimlanes of change.
Real change moves in four swimlanes, and most organizations only ever run one of them. Picture someone in the water who needs to reach a boat.
Stakeholder management is understanding how far each person is from the boat to begin with. Not everyone starts in the same place, and pretending they do is the first failure.
Communications simply creates awareness that there is a boat. That’s all it does. It’s necessary, but it’s the beginning of the work, not the work itself.
Transition management is what happens once you’re on the boat — what’s actually new, what’s different, what the job looks like now. This is the swimlane no one understood back then, and almost no one understands now.
Training teaches people how to swim to the boat. It’s the skill-building, and it only works if the first three lanes have already done their job.
Organizations love communications and training because they’re visible and easy to schedule. Stakeholder management and transition management are where the real difficulty lives, and they’re the ones that get skipped.
Back then, at least we knew what the job was.
With a migration like PeopleSoft to Workday, the delta was bounded. People were learning to process payroll in a more efficient system. Recruiting in a new system. Maybe finding some incremental improvements along the way. But no one was reinventing payroll or recruiting. There was a job, and everyone knew what the job to do was.
AI is different in kind, not just degree. We’ve handed everyone the tool and told them to use it in everything — to be more efficient, to be AI-first. But we don’t even know what the job is anymore. AI doesn’tjust move the work to a better system. It lets us blow the work up and rethink it entirely. The delta isn’t bounded. It’s open-ended.
And there’s one more thing that’s changed. Back then, there were process owners. They were the people who understood the work deeply enough to translate the delta for everyone else — to say, here’swhat’s different about your day tomorrow. With AI, everyone has access to the tool, but no one is playing the translator role. The person who used to define what “different” means is gone, and we never replaced them.
Communications in the absence of details creates anxiety.
That’s the whole problem in one sentence. You’re telling people they need to do this, work it in, be more efficient, be AI-first — with no idea of what good actually looks like. The message arrives, but the substance behind it doesn’t. And in that gap, people don’t get motivated. They get anxious.
No one knows what good looks like.
Not the employee being told to use AI. Often not the leader doing the telling. The mandate is real, the urgency is real, but the picture of the destination is missing. You can’t communicate your way out of that. No amount of messaging fills a hole where the substance should be.
The work is in the delta, function by function.
This is where it comes back to doing the work before the words. The way through AI adoption isn’t a louder rollout email. It’s going function by function — finance, recruiting, customer service, legal — and doing the unglamorous work of figuring out which tasks AI can actually handle well, and tackling them in digestible chunks.
I’ve seen this work. A recruiting team builds a system that quickly generates job descriptions in the right format, ready to drop straight into the applicant system. A divisional communications team teaches AI to produce a strong first draft of a leader’s monthly note to their division. In both cases it worked because someone did the work first — and because once it was tested, perfected, and proven, it could be packaged and shared with everyone else. That’s the translator role, rebuilt deliberately.
That’s how you rebuild the delta. You define what tomorrow looks like for a specific team doing specific work, so that when you finally do communicate, you’re describing a real destination instead of a vague aspiration. You’re replacing the process owner who used to translate for you — deliberately, by function — instead of pretending the translation will happen on its own.
AI didn’t invent a new problem. It exposed an old one we never solved: we keep mistaking communications for change. The message was always the easy part. The hard part — understanding the delta and articulating it honestly — is exactly the work most organizations still skip.
It’s the work before the words. It was true in the PeopleSoft-to-Workday era, and it’s true now. If your organization is going AI-first and no one can yet say what that means for the people doing the work, let’s talk about closing that gap.
