Many businesses start with the wrong AI question: Which software should we buy? A better question is: Where do we regularly lose time during the day?

Maybe enquiries are copied from one place to another. Maybe call notes are summarised by hand. Maybe someone answers the same five questions every day. That is a good place to begin.

The first step does not have to be perfect. It only needs to be small enough to test and useful enough to make a difference.