Is AI coming for restaurant jobs? An operator's answer

Somebody in every room asks it. Sometimes out loud, usually not. It is the question sitting underneath every other question about AI in a restaurant, and it deserves a straight answer rather than reassurance.

Here is mine, from somebody who has spent a career on the operations side of stores.

What it is not going to do

It is not going to cook. It is not going to run a pass on a Friday, notice that table twelve has been waiting too long, or talk a guest down after a wrong order. It is not going to pick up a shift when someone calls out at four o'clock.

The physical, social, in-the-moment part of the work is most of the work, and nothing currently being sold to restaurants comes close to it.

What it does take

Paperwork. The office side of the job.

Keying invoices into food cost. Rebuilding the Monday report out of three systems. Drafting next week's schedule. Writing the fortieth reply to the fortieth review. Reconciling a count that did not match.

That work lands mostly on managers, and it is the reason a general manager spends hours in the back office that were supposed to be spent on the floor. Taking it away does not remove the manager. It gives the manager back.

The honest part

Some people's whole job is that paperwork.

In a group, there is often one person whose week is keying invoices, or one person whose job is building reports. If most of that gets automated, that job changes, and pretending otherwise would be the kind of answer that makes people stop trusting everything else you say.

What that person does next is a real decision, and it belongs to the operator, not to a piece of software. In my experience the person who has been keying every invoice for three years is the one who knows which vendor short-ships, which credit never arrives and which price crept up in March. That knowledge is exactly what you want checking the automated version. Moving them from typing to checking is usually the better use of them.

I will also say plainly: I turn down work where the point is cutting people. Not because it is impossible, but because it is not the problem most restaurants have.

Most restaurants are short of people, not long on them

The usual problem in this industry is the opposite of too many staff. It is open shifts, managers covering, and good people leaving because they are doing two jobs.

Hours freed from the office in that situation do not become layoffs. They become a manager who is on the floor at the dinner rush, a schedule that goes out on Thursday instead of Sunday night, and fewer people burning out.

Ordering technology is a different conversation

Kiosks, online ordering and QR menus have genuinely changed front-of-house work, especially in quick service. That is worth being honest about too. But it is not what most people mean by AI, and it is a decision that was already being made long before anybody was talking about language models.

Lumping the two together makes both conversations worse.

What to tell your team

Tell them before they guess. Silence is what makes people assume the worst.

Be specific about what is changing: this report stops being built by hand, these invoices stop being keyed in, this draft schedule gets made before you start. Be specific about what is not changing: who runs the shift, who makes the call, who owns the schedule.

Then involve the people who do the work now. The fastest way to build trust is to start with the jobs they already want gone, because nobody feels threatened by losing a task they hate.

---

Wondering which of your team's hours are going to paperwork? Take the free check — eight questions, no email, and it names the two worth taking back first.