Most of my career, the change playbook never let me down.
ERP rollout, restructure, system migration. Same spine every time, and it worked.
Then I watched teams run an AI rollout the same way, and the playbook that never failed me suddenly didn't fit.
A new survey of 1,000 change leaders, run by Dan Schawbel's Workplace Intelligence with Deloitte and reported in Inc., found that 53% use the same approach for every change, whatever the type or audience.
For most of what we do, that is fine. A restructure and an ERP go-live are different, but not different in kind.
AI is different in kind. Here is where it breaks.
Start with the go-live. Every ERP has one. You switch the old system off, and on Monday morning people have no choice but to work the new way. Most of us assume AI works the same. It has a launch date on the plan, so we treat it like a cutover.
It isn't one. On paper, there is a go-live. In real life, nothing gets switched off. The old way still works. People can sit through every session, nod in every demo, and keep working exactly as they did before. Adoption you cannot enforce is a different problem, and the ERP playbook has no play for it.
Then there is the work itself. An ERP asks you to do a known task in a new place. AI asks you to trust output you can't fully verify, to judge when not to use it at all, and to sit with the quiet worry that the skill you built your career on is suddenly worth less.
None of that is a training gap you can close with another session. It is a different set of frictions, and the ERP playbook was never built to see them.
You cannot run an AI implementation like an ERP implementation.
The old playbook isn't wrong. It was built for a change that ends. This one doesn't.