Related question
Why do our best people resist moving to the company-standard AI platform?
Short answer
Because they built something. Your most capable AI users spent months training a tool to match their work style, standards, and responsibilities. Their resistance is not change aversion. It is personal investment in something that works, and they know it works.
Not everyone resisting your platform transition is a weak user hiding behind a familiar tool. Some of your most competent people have done the opposite. They mastered prompting and then went further, spending months teaching their preferred tool how to behave, shaping its output to match how they work.
The customizations sound small individually. Someone may have trained their tool never to use certain punctuation, or to always structure a response a particular way, or to know the standing context of their role. Multiply that across a year of daily use and hundreds of deliberate choices, and you have a highly tuned workflow running quietly in the background every time they open the tool.
Those people need something different from standard training. Training assumes a skill gap and they do not have one. What they need is help deconstructing what they built in the old platform and reconstructing something equivalent in the new one. That is different work and it takes time.
Organizations serious about a successful transition give these individuals reduced responsibilities during the migration, or pair them with someone experienced at moving between platforms. Without that support, you will either lose the productivity of your most capable users or lose the people themselves.
For everyone else in the migration, three things reduce the friction substantially. Explain the business reason for standardizing, because security, cost, support, and governance are legitimate reasons and people accept legitimate reasons. Train on real work rather than generic demonstrations, and teach the underlying skill rather than the specific product. And allow for exceptions where a specialized use case genuinely justifies one. The objective is not identical behavior across the company. It is enough consistency to govern the environment without destroying productivity.
Identifying who falls into this category is the first step. Treating them differently from everyone else is the second.