AI in restaurants: why guests only notice it when it goes wrong

AI in restaurants got a public face in the Netherlands when BUNKR opened the country's first AI restaurant. Television, radio and newspapers were all over it. Part of the public openly hoped the place would close within a month.

That reaction says more about guests than about technology. Robin Vromans of BUNKR put it this way: the responses went left or right, with nothing in the middle. Guests who called it "a stupid idea", and industry colleagues who mostly wanted to know what you could learn from it. If you run two or three restaurants, you'll recognise that tension. You can see AI saving you time, but you don't want to be the restaurant guests complain about because a robot ruined their evening.

So you wait. Meanwhile, last month's reviews go unanswered, your Google profile hasn't been touched in six months and nobody asks happy guests for a review. That's a choice too, just one you don't see. A competitor three streets away who pulls in five new reviews every week is quietly overtaking you on Google Maps. The real risk isn't using AI. It's using AI in the wrong place, or not at all.

Guests want people, Google wants speed

The pushback BUNKR got was about technology at the table. Guests come for hospitality, for someone who explains the wine and notices the kid needs an extra napkin. They don't want a screen in the way. But that same guest reads twenty Google reviews beforehand, checks whether the owner responds and looks up your opening hours on your profile. That part of the journey is digital, and there they expect speed and complete information.

That's where AI belongs: in the work that piles up after closing time. Replying to a review within a day instead of three weeks. Asking a guest for feedback on WhatsApp an hour after their visit, while the experience is still fresh. Catching a bad experience privately before it ends up public on TripAdvisor. Nobody at the table notices an agent at work. Your ranking does.

Keep front of house and back office separate

We use a simple framework with restaurants that are on the fence. Split every process into front and back. The front is everything guests experience during their visit: the welcome, the service, the plate. The back is everything before and after: discoverability, reviews, the website, follow-up. With businesses running one to ten locations, we see the same pattern again and again: the front gets all the attention, while the back runs on whatever time the owner has left on a Sunday evening.

Four agents for the back office

For the back office, we build AI agents that each take over one job. One automatically asks for reviews by email, SMS or WhatsApp. One replies to reviews on Google, TripAdvisor, Booking and Facebook in your tone of voice. One creates and optimises posts for your Google Business Profile. And one builds and maintains your website, so your menu and opening hours are always right. The better the back office runs, the more time you have for the front of house.

Why fellow operators mostly ask questions

The curiosity BUNKR sparked among industry peers is telling. They asked what lessons there were for the regular way of running a restaurant. The answer is less spectacular than an AI restaurant: start where nobody sees it. An owner with four locations who has to reply to dozens of reviews every week wins back hours that way, without a single guest having to put up with a robot.

AI in restaurants doesn't have to be a publicity stunt that splits opinion down the middle. The restaurants getting the most out of it right now use it where guests don't see it and Google rewards it. AfterTheTable builds exactly those agents for reviews, your Google profile and your website, so your team can focus on what guests actually want: people at the table.