Burger King plans an AI coach for staff — but governance, accuracy should come first, say experts
AI is transforming workplaces everywhere — and it may be playing a role in getting your burger.
Burger King says it plans to bring an AI-powered voice assistant called Patty to Canadian restaurants in the second half of 2026, delivering real-time prompts through the headsets used by frontline staff. The tool is designed to listen to employee-customer conversations and help with tasks such as order assembly and upselling, while also flagging operational issues like out-of-stock items or washrooms that need cleaning, according to the Canadian Press.
The company has described this approach as coaching rather than tracking individuals, saying it’s not about “scoring individuals” or enforcing scripts. For HR leaders, the strategic question isn’t whether always-on coaching of employees is technically possible. It’s whether a constant “voice in the ear” can be implemented in a way that improves productivity and performance without eroding trust, psychological safety, and equity, says Valerio De Stefano, a professor at Osgoode Hall Law School at York University.
Transformation starts with strategy, not feasibility
De Stefano believes that organizations can slide into the wrong decision-making sequence when a tool looks cheap and easy to deploy. “This is a typical example of a situation in which technology is driving managerial choices, rather than managerial choices guiding the use of technology,” he says. “The mere fact that something is technically feasible and can be implemented at relatively low cost, doesn’t mean that it should be introduced, nor that it will have positive effects.”
In practical HR terms, that means a pilot with new technology should begin with a sharply defined purpose statement that a frontline employee would recognize as legitimate: reducing time-to-competence, improving safety, supporting multilingual onboarding, or decreasing the cognitive load during peak periods — but it also means setting a boundary on what the tool is not for, especially where real-time prompts could quietly become evidence in performance management, according to De Stefano.
One aspect of Burger King’s new AI tool test that may catch the attention of leadership is that the system’s value depends on listening to work in progress, and on turning that work into data that can be reviewed. Whether employees experience that as “coaching” or “surveillance” will be determined by who sees the information, how long it’s retained, and what decisions it can influence, according to Vanesa Cotlar, Vice President People and Culture at PolicyMe.
The distinction between helpful coach and digital surveillance is based on how the tool is positioned and how the information collected is used, according to Cotlar. “For example, if the user of the headset clearly understands what’s being collected, they see the feedback first, and it’s framed as ‘here’s something that can help you,’ not ‘here’s a record that might be used against you later,’ that's coaching,” she says. “It tips towards surveillance when the data is primarily flowing up to dashboards that monitor, rank, and discipline people — especially if that wasn’t crystal clear on day one — so organizations that manage the line well and are honest about intent, have clear information on data use, and let humans challenge the tool when it doesn’t reflect the reality they were promised.”
Redesigning how decisions are made on the floor
For HR leaders, “co-work” implies co-design, so if a tool will prompt behaviours, then frontline workers, supervisors, and people managers should help define what good service looks like in their setting, what types of prompts are useful in real conditions, and what happens when the tool is wrong, says Cotlar.
“It’s important to consider that this is a big redesign in how decisions are made, especially on a floor like Burger King's,” she says. “CEOs and CHROs should be asking, ‘How do we bring the humans who will be using these tools into the development of these experiments? How do we ensure the tools add real value to both our businesses and the humans using them?’ — it should be about collaborative design to build new systems that actually work.”
Continuous voice-based coaching also raises privacy questions, as even if a system is positioned as live prompting, it may still involve analyzing voice characteristics and create new sensitivity around access, retention, and secondary uses of data. “These systems may involve the collection and processing of biometric data, since they analyze features such as tone, cadence, and timbre of voice, as well as other potentially sensitive information,” says De Stefano. “That kind of data collection seems disproportionate to the employer’s legitimate needs, especially in a setting [like Burger King].”
For Canadian organizations, the strategic safeguard is to treat a constant-coaching pilot like a high-impact data initiative, in which HR should be able to explain, in plain language, what’s captured or inferred, what’s stored, who can access it, how long it’s kept, and what it’s explicitly banned from being used for, says Cotlar.
Coaching, benchmarking the right targets
De Stefano says that voice-monitoring tools can also produce uneven or inaccurate outcomes if the benchmark for “good” speech or tone doesn’t match the workforce, particularly in environments with diverse accents and speech patterns. “Voice-monitoring tools may misinterpret workers whose speech does not align with the benchmark on which the system has been trained,” he says. “This can be particularly problematic for workers with certain accents or speech patterns, including those linked to race, ethnicity, gender, disability, or other protected characteristics — as a result, these tools may create significant discrimination risks, with workers from more vulnerable groups being disproportionately flagged as not meeting the algorithm’s standards.”
That risk changes how HR should evaluate performance impact. Beyond measuring speed of service or training time, leaders need to monitor who is corrected most often, who is flagged as missing courtesy language, and whether false positives cluster by location, shift, or role, says De Stefano, so the tool is designed to standardize quality instead of bias.
Burger King’s stated intent is to expand Patty eventually across many of the roughly 380 locations in Canada. The prospect of fast scaling is exactly why HR leaders should set governance before a pilot becomes a default, says Cotlar, who draws a bright line between a tool that supports the worker and a tool that primarily supports monitoring. “It tips towards surveillance when the data is primarily flowing up to dashboards that monitor, rank, and discipline people — especially if that wasn’t crystal clear on day one.”
‘The future is here’ for coaching frontline work
Cotlar frames the moment as a shift to real-time human-machine collaboration that leaders need to govern deliberately.
“The Burger King example shows that the future is here and we need to learn how to co-work with machines in real time — it isn't about if, it’s about when,” she says. “I’d love to see CEOs and senior HR leaders write down a clear peoplefirst AI philosophy, invest as much in communication and codesign as they do in the tech itself, and build real feedback loops so workers can say, ‘This isn’t working for us,’ and actually see the organization adjust — that’s how we build AI that makes frontline work better, not worse.”
