Is the Best AI in Restaurants Boring AI?

While ChatGPT ordering launches from Starbucks and Little Caesars made headlines earlier this year, the best AI in restaurants is boring AI, said Santiago Noziglia, CEO of Globant's Retail, CPG & Automotive AI Studio.

“The guest just feels it: the line moves, the order's right, the food's ready when promised. The real transformation comes when the restaurant gets sharper without losing its human feel. Fewer stockouts, fewer surprises, better timing, less friction for staff and guests.”

Restaurants are becoming more responsive in ways guests can actually feel, even if they never see the AI itself with better labor planning, smarter inventory, tighter prep timing, and smoother order flow, Noziglia added. All  of that is happening in real time based on what's going on in that specific location, at that specific hour.

If the tool helps with discovery, simplification, or planning, that’s interesting. If it adds a step between the guest and a simple reorder, it misses the point.

The Starbucks and Little Caesars examples are useful signals that show major brands actively testing where AI can improve the guest experience, and that matters, Noziglia said, adding that the most impactful AI won't feel like AI, it'll just make things faster, easier, and more reliable and keeps people coming back.

“The question I always come back to is what problem is it actually solving? If the tool helps with discovery, simplification, or planning, that’s interesting. If it adds a step between the guest and a simple reorder, it misses the point. The brands that should be paying attention are the ones asking whether their AI is connected to real operations: inventory, menu availability, order flow.”

Additionally, there are risks of using AI in high guest touchpoints, particularly putting it in front of guests before issues are fixed what's behind the counter, Noziglia noted. For example, a drive-thru bot that doesn't know you're out of a menu item or that the kitchen is backed up six orders deep isn't helping; instead, it's actually creating a new problem. 

“And in QSR, one bad experience gets talked about in the car ride home or posted online that night. Drive-thrus, pickup counters, kiosks are all high-stakes moments where people expect speed, accuracy, and a clean handoff. If AI adds friction or gets the order wrong, guests feel it instantly. The real risk isn't AI failing. It's brands using AI as a front-end band-aid when the actual issue is operational. If the back end is disconnected, no amount of conversational AI at the speaker box will save you.”

Brands should use AI where it drives accuracy, speed, and consistency such as in forecasting, staffing, inventory visibility, order accuracy,  said Noziglia.

“Trust comes from keeping the experience predictable and honest. That's it. No magic. When something goes wrong (and it typically will), there still needs to be a human path to make it right, fast.”

Guests care about three things: did you fix it, did you own it, and is next time better? AI supports that loop, but it doesn't replace accountability. Brands that understand that will win.

AI can help by catching problems sooner, accelerating recovery, and making responses more consistent, but the real repair happens in follow-through, according to Noziglia. 

“Guests care about three things: did you fix it, did you own it, and is next time better? AI supports that loop, but it doesn't replace accountability. Brands that understand that will win.”

In order for AI to be used advantageously, it has to plug into the systems that actually run the store including POS, inventory, labor, kitchen ops, because if it only sits on top of the experience layer, it can't do much and is just a decoration, Noziglia said. Legacy POS can be challenging because they hold critical transaction data, but most weren't built to talk to modern tools.

“The real job is building clean, real-time integrations and not just stacking another layer of tech on top. AI needs to know what's happening in the store as it happens, so it helps teams make better decisions instead of creating more work. The litmus test is simple: if your store manager has to do extra work to make two systems talk, adoption dies. That burden kills more AI projects than bad algorithms ever will.”

AI raises the trust bar, Noziglia said. 

“When a person gets your order wrong, it feels like a human slip. When AI does it, it feels like the brand should have known better. A wrong modifier, a missed allergy note, a bad pickup time, a bot recommending something that's out of stock — any of those can break trust fast.”

But he anticipates that tolerance will shift as AI will increasingly do its best work out of sight in functions such as adjusting prep timing, flagging inventory gaps before they hit the guest, and  optimizing staffing before the lunch rush even starts. 

“The more AI quietly makes things work, the less people will focus on the occasional miss. Reliability earns patience,” Noziglia said.