When I first heard about Burger King’s new AI headset system, I assumed it was one of those tech stories that sounds more dramatic than it is. I went and did the research. It was not more dramatic than it sounds.

Burger King’s parent company, Restaurant Brands International, confirmed in late February 2026 that it is piloting an OpenAI-powered voice assistant called Patty inside employee headsets at 500 US restaurants. The plan is to expand Patty to all US Burger King locations by the end of 2026. Patty listens to drive-through conversations, answers operational questions from staff, alerts managers to low inventory and broken equipment, and tracks whether employees are saying keywords like “welcome,” “please,” and “thank you.”

They call it a coaching tool. That’s the word I want to examine.

What Patty Actually Does

Let’s be fair to the parts of this that are reasonable. An AI that can tell a manager when the Diet Coke machine is running low: fine. An AI that can answer a new employee’s question about how many bacon strips go in a specific sandwich without them having to find a supervisor: actually quite useful. Operational intelligence, real-time information, reducing the friction that makes frontline work unnecessarily hard. That part of Patty, I have no real issue with.

The part that I do have an issue with is listening to drive-through conversations and generating a friendliness score based on whether the employee said the right words.

Burger King’s chief digital officer Thibault Roux was careful in interviews to say this is not about scoring individuals, it’s about restaurant-level coaching and service patterns. I’ll take him at his word on that, for now. But I’d also ask him to think about what happens when that data is collected consistently over months. When franchise owners start looking at it. When it starts appearing in performance reviews, even informally. When an employee who doesn’t hit the keyword frequency starts getting conversations pulled out and shown to them. History tells us that metrics that get collected tend to get used.

Sources: Fortune (27 Feb 2026), NBC News (26 Feb 2026), Fast Company (27 Feb 2026)

The Specific Thing That Makes This Different

I’ve spent a lot of time at Techosaurus helping businesses use AI to do more, to automate the repetitive, to free people up for the things that actually need a human in them. That’s the frame I bring to all of this. AI as amplifier. AI as assistant. AI that makes the people it works alongside better at their jobs.

Patty isn’t doing that. Patty is doing the opposite. It’s taking a fundamentally human interaction, someone saying hello to a customer and taking their order, and putting an AI in their ear during that interaction to evaluate their warmth. If you have to be reminded by a machine to say please and thank you, you’ve already removed most of what makes a human interaction worth having.

I’m not being romantic about hospitality work. I know it’s hard, I know it’s fast-paced, and I know that not every shift goes the way it should. But the idea that you can improve customer warmth by training people to say specific keywords, which an AI will then validate, is to completely misunderstand what warmth is. Warmth is not a list of words. It’s tone, timing, attentiveness, recognising when someone’s having a rough day, responding to what’s actually in front of you. You can’t score that with keyword detection.

What Burger King have actually built is a system that will train people to perform friendliness rather than practise it. That is a very different thing.

We’ve Seen This Before

McDonald’s trialled AI voice ordering at their drive-throughs in partnership with IBM and ended that partnership in 2024, largely because the AI was getting orders badly wrong. It wasn’t ready. The lesson from that wasn’t “try a different AI in a different part of the experience.” The lesson was that there are places in the fast food operation where AI genuinely helps and places where it doesn’t, and working out which is which requires some honesty about what the technology can and can’t do.

I said on Prompt Fiction that there are too many companies trying to shoehorn AI into places it doesn’t belong. The drive-through order window is an interesting case. Yes, you could replace it with a touchscreen. Yes, you could replace it with a QR code that people scan from their car. Both of those would work better than an AI voice system, and they’d probably be cheaper and more reliable. What you can’t replace is the person who takes your order, recognises a regular customer, remembers they always get extra pickles, and has a small genuine moment of human contact in an otherwise automated experience. If you’re going to keep a person in that role, the last thing you should do is put an AI in their ear telling them to smile more.

Where the Line Is

None of this is an argument against AI in retail or hospitality. There are genuinely good uses. Smarter inventory systems. Better scheduling tools. Training platforms that help new employees get up to speed faster. Predictive maintenance for kitchen equipment. Analytics that help managers understand peak demand. All of those make the operation better without touching the human interaction at the heart of it.

The principle I keep coming back to is the one I use at Techosaurus: AI should amplify what people are already good at, not audit what they do when they’re already doing it. There is a meaningful difference between giving someone better tools and watching them use the tools they already have. Patty sits on the wrong side of that line.

I hope Burger King’s pilot finds that too, and that the data shows what I suspect it will: that the metric tracking does nothing for warmth and something for anxiety among the staff who know they’re being monitored. I’d be very happy to be wrong. But I’ll be watching this one.

I discussed this topic on the latest episode of Prompt Fiction. Listen to Chapter 12, Part 1 here.

Scott Quilter | Co-Founder & Chief AI & Innovation Officer, Techosaurus LTD