The best migration is the one nobody notices
If people notice your migration, something went wrong. Good change is invisible to the business it serves.
Nobody in a business cares where their mailbox is hosted. They care that email works on Monday morning, that their files are where they left them, and that they can get on with their day. Moving a company from one system to another is a big event for the person doing it. For everyone else, it should be a non-event.
That's the standard: the only thing users should notice afterwards is that things work as well as before, or better.
The work happens before the switch
Learn the process until there are no steps you don't understand. Ask people who have done it before, because their mistakes are cheaper to learn from than your own. Take a full backup you hope never to use. Pick a window when nobody is working, and tell people about it anyway, so nobody is caught mid-task.
I have moved a company's entire Microsoft 365 tenant this way, overnight. It was my first one. There was no downtime and no data lost, and the next morning staff simply carried on working. My announcement was one message saying it was done.
A calm migration night is earned before it starts.
Cheaper is not the same as cutting corners
The consultants' quote for that migration was about a hundred dollars per gigabyte. Once you understand it properly, you can do the same work at the same quality for a fraction of the cost. Saving money done right means paying only for what you actually need.
What this has to do with AI
A lot of companies are approaching AI the way people approach an expensive migration: under pressure, because a competitor did it, with a consultant's quote on the table. The lesson carries over.
Don't follow the trend, and don't copy what your competitors are adopting. Start with an honest question: do we even need AI here? Then understand your business processes properly: who does what, where the time goes, where the money goes. Only then decide where AI fits, and plan it so the work never stops while it comes in.
Transformation is not the dramatic launch. It is the preparation that makes the switch feel ordinary. That is the kind of applied AI work I want to do.