The ‘Please and Thank You’ Police: Why Burger King’s AI is Listening to Your Conversations

The drive-thru has an almost cinematic quality. A disembodied voice tells you to order a Whopper while you idle up and roll down the window. It is routine, unremarkable, and wholly American. However, Burger King has subtly added a new layer to that conversation: “Patty,” an artificial intelligence that rides inside employees’ headsets and listens to both what is ordered and how the conversation goes.

The system is a component of a larger platform called BK Assistant, which has been implemented in about 500 US restaurants thus far and is based on an OpenAI base model. The pitch seems harmless enough on paper: Patty assists employees in remembering ingredients, notifies managers when supplies run out, and presumably notifies staff when the restroom needs to be cleaned. However, the part of the announcement where the AI listens for words like “welcome,” “please,” and “thank you” as indicators of what Burger King refers to as “friendliness” seemed to make people’s hair stand on end.

Burger King’s chief digital officer, Thibault Roux, characterized it as a coaching mechanism as opposed to a monitoring tool. “It’s really a coaching tool,” he told NBC News, presenting the AI as a tool to make workers friendlier. The company was quick to add that BK Assistant won’t be listening to every conversation employees have and isn’t designed to score individuals or push anyone to stick to a rigid script. However, the moment a customer pulls up to place an order, the system does activate, and it doesn’t stop until the car leaves.

It’s difficult to ignore how this framing—”coaching, not scoring”—has evolved into the standard language used by businesses whenever a new monitoring tool raises concerns. The real workings are fascinating: a headset with a microphone, an AI model that analyzes conversational cues, and a database that gathers information on service “patterns.” Depending on which side of the headset you are on, that may or may not qualify as support or surveillance.

Online, there was a quick and direct backlash. The rollout was dubbed “gross” by social media users, who also referred to it as “peak late-stage corporate behavior,” a term that often sticks. Some of that response might be reflexive; people don’t like hearing that their place of employment has ears these days, even if it’s just hypothetical. However, there is also a plausible argument that the worry isn’t excessive. Employees have always chosen to use the exaggerated “customer service voice,” which has long been ridiculed in online memes as being drastically different from how people actually speak. A different energy is added to that performance when an algorithm listens for compliance.

Burger King is not the first fast-food business to experiment with AI-powered operations. Before ending the project last year, McDonald‘s tested automated AI voice ordering at more than 100 drive-thrus. The company cited enough documented errors, the majority of which were enthusiastically posted on social media, to make further expansion unfeasible. Popeyes, Arby’s, and Wendy’s have all been conducting their own experiments. Even though no one has yet found the ideal version, it is evident that the industry is aiming for something. The Associated Press reported that there has been mixed success.

The employee-facing angle is what sets Burger King‘s strategy apart. Automated voice systems that take orders at the speaker box have been the focus of the majority of well-known AI experiments in fast food. The positioning of BK Assistant is different. Instead of focusing on the customer order itself, it is integrated into the workflow, inside the headset, and targets staff behavior and operations. That is a significant difference in terms of both optics and function.

There’s a sense that Burger King is genuinely trying to solve something real here — the complexity of fast-food operations isn’t trivial, and a tool that reminds workers about limited-time ingredients or flags menu availability issues in real time could be legitimately useful. The question is whether it was necessary to add a “friendliness” monitoring feature to that same tool or if it’s just the kind of feature that makes sense in a product meeting but doesn’t work well in practice. Roux admitted that the company revised its definition of friendliness, implying that this wasn’t a clear-cut design decision.

The full US rollout is expected by the end of 2026, which gives Burger King — and the rest of the industry — time to refine what these systems actually do and how they’re communicated to the people using them. It’s another matter entirely whether that’s sufficient time to foster true trust among staff members. The AI Patty might never say “please” on her own. But when you do, it’s undoubtedly listening.