Burger King, Starbucks enlist AI ‘assistants’ to memorize recipes and check inventory — but it's also started tattling on workers

The integration is part of a trend of AI growth in the struggling quick-service and fast-food restaurant sector. They're designed for everything from inventory management to metrics to predicting customer orders (2) to shift scheduling — with particularly negative effects on workers.

But not all tools are the same, nor do they have the same impacts, even though they're all generally described as the business "adopting AI."

A sector under strain

Fast-food restaurants are facing a financial crunch as tastes change and budget-conscious consumers make more food at home, according to Axios reporting (3).

Starbucks in particular is coming off of a rocky few years. After running through several CEOs, closing hundreds of stores, slashing corporate staff and trimming the menu, its sales and profits are finally rebounding this year (4).

The chain introduced its tablet-based AI barista assistant, Green Dot Assist, in 2025 amid its turnaround strategy and renewed focus on customer service (5). The tool aims to help workers with "streamlining operations" and "reducing friction," while giving them more time to spend making the drinks and talking to customers, the company claims (6).

Burger King has a similar headset-based AI tool, dubbed Patty, which attracted significantly more attention when it was rolled out — and not all of it positive.

According to reporting from The Verge, Patty can answer verbal questions about how to make common items or clean the ice-cream machine. It's also integrated with the company's point-of-sale system, meaning it can send an alert when a machine is down for maintenance or an item is out of stock (7).

However, more concerningly, because it's voice-based, it's listening. Although Burger King stated for the record that the system, which uses Open AI technology, does not "record conversations or evaluate individual employees," according to BBC reporting (8), it is trained to pick up on phrases like "please," "thank you" and, "Welcome to Burger King," assigning a "friendliness score" that is reported back to managers.

And this isn't the only example of AI-led employee surveillance; in April, "frontier AI" startup Andon Labs launched the first ever AI-run store, where Luna, the store's AI manager, analyzed security camera footage and noted an employee using a smartphone during work hours. Luna subsequently updated the employee handbook to set stricture rules for employees, signalling where AI work surveillance may go next.

Meanwhile, Wendy's, McDonald's and Taco Bell have experimented with having a chatbot take drive-thru customers' orders directly (9), with mixed and sometimes humorous results (10).

Understanding AI claims

As a consumer, it's easy to get overwhelmed by stories about AI, whether as an economic miracle or job killer.

It's also easy to fall into the trap of thinking and talking about AI as if it's one single thing, rather than a patchwork of different tools with different benefits, limitations, impacts and underlying technologies.

Let's take Starbuck's new AI barista buddy, Green Dot, as an example. What is it, exactly?

It was made with Azure, Microsoft's cloud computing platform. It uses a large-language model (LLM) from OpenAI, just like the one underlying ChatGPT, but it has been fine-tuned to respond to coffee and Starbucks-related questions (11). Workers can ask it questions in conversational language. It's rigged up to pull information from Starbucks's internal resources when possible; a process called grounding that helps LLMs generate more accurate text (12). It runs on iPads set up at workstations in stores (13).

So, to simplify: It's a souped-up version of ChatGPT loaded onto an iPad.

A mixed bag for employees

Starbucks COO Mike Grams told Business Insider that Green Dot saves time by preventing workers from "fumbling through a computer" or glancing at laminated recipes taped inside the counter.

Business Insider spoke to several Starbucks staff members and the picture was more mixed, with many of the complaints echoing those of other employees who've been told to use LLM-based tools at work. One San Francisco-based worker said it was "hit or miss" and could take 10-20 seconds or more to answer questions. Another complained it would frequently say "I don't know" and returned inconsistent and sometimes incorrect answers.

Posts around the web (such as on TikTok (14) and Reddit (15)) about the tool's rollout have attracted many comments from employees expressing doubt that the chatbot is a significant improvement over the old system, which combined an internal intranet (called "Store Resources") with at-a-glace physical recipe cards.

In fact, one barista that Business Insider spoke to said Starbucks changed how internal information is organized to accommodate the bot, making it harder to look up answers manually.