I have sat on enough hotel-tech panels this year to notice a change in the questions. Nobody asks whether to use AI anymore. Everyone asks which tool to buy.
It is the wrong first question.
What happens when the tool comes first
If you don't understand how your teams actually operate, where the time is lost, and how decisions really get made, adding AI won't fix anything. It layers complexity on top of the problems you already have. Now you have the old process, plus a subscription, plus a new system somebody has to feed.
I watched this from the inside for a decade. The pattern is always the same. The tool gets bought because a board member asked about AI. It gets piloted by whoever had capacity that month. Six months later it is a line in the software budget that nobody defends with much conviction.
The tool was never the problem. The operation was never examined.
Start with the work
Before any tool, sit with the actual work. Where do the hours go in a week at reception, in reservations, in finance? Which reports get built by hand every Monday? Which decisions are still being made on instinct because the numbers arrive too late to matter?
That exercise is boring. It is also where the money is. When you start with the process, the right automation becomes obvious, and often it is smaller and cheaper than the platform you were about to buy. When you start with the tool, it becomes noise.
This is not a project you can delegate
The other mistake is structural. AI gets handed to an innovation team or a keen manager, and leadership waits for results.
Meaningful change in how a company works does not happen in a side project. It changes how decisions are made and what people spend their day on, and that is leadership territory. If the top of the company is not personally in it, the organisation reads the signal correctly and carries on as before.
The winners in this industry will not be the companies with the most tools. They will be the ones where the way of working actually changed, starting at the top.
Buy the tool last.