0% read
    Skip to content
    Monday, November 9, 2026
    Partner with us
    Retail News AI
    Technology

    Jabra Bets on Voice as the Gateway to AI for Retail's Frontline Workers

    TL;DR

    Jabra VP Andrew Doyle argues that voice is becoming the primary interface between retail frontline workers and AI agents, making headset audio quality, noise cancellation and interoperability decisive for whether retail AI delivers real productivity.

    Key Takeaways
    • 1Voice is emerging as the natural frontline interface for retail AI: Jabra cites speaking speeds of 200 to 300 words per minute against typing at 40 to 70, so associates can query AI agents heads-up and hands-free.
    • 2Audio quality becomes a data problem once AI listens: Jabra claims around 96% transcription accuracy for enterprise-grade hardware against 64% for consumer headsets, and 99% microphone noise cancellation on the Perform 75.
    • 3Interoperability and shift-work design matter as much as intelligence: hot-swappable batteries, 30-hour runtime, tap-to-pair Bluetooth and integration with Zebra Workcloud Sync and Teams Walkie Talkie decide whether deployments stick.
    By Retailnews.ai EditorialAug 17, 202611 min read
    Share:
    AI-Assisted Content

    As retailers invest heavily in AI agents, workforce apps and connected devices, Jabra believes one critical part of the technology stack risks being overlooked: the audio endpoint. Andrew Doyle, VP for Frontline Workers at Jabra, explains why accurate voice communication, noise cancellation, interoperability and hardware designed around the realities of shift work could determine whether AI delivers genuine productivity gains on the shop floor.

    Retail's AI revolution is usually discussed in terms of models, software and data.

    But what happens when an employee standing in a noisy stockroom actually needs to talk to that AI?

    Retail reportsWant the bigger picture? Read our monthly Retail Intelligence reports, with the data behind the headlines.

    For Andrew Doyle, VP for Frontline Workers at Jabra, that question could become increasingly important as retailers move artificial intelligence from corporate offices onto the shop floor.

    Speaking with Alex Rezvan, Founder of The Retail Podcast, Doyle argued that voice is emerging as a natural interface between frontline employees and AI agents, and that the quality of the hardware connecting the two could ultimately determine whether those systems work in practice.

    It is a deceptively simple proposition.

    Retailers can invest millions in sophisticated AI platforms capable of guiding associates, retrieving information and triggering workflows. But if an employee cannot clearly hear the system, or the system cannot accurately understand the employee amid the noise of a busy store, the intelligence behind it becomes considerably less useful.

    That puts an established technology in a potentially new strategic position.

    The headset could become the gateway between the frontline worker and AI.

    Jabra Is Taking Its Audio Expertise to the Frontline

    Jabra is part of Danish technology group GN, or GN Store Nord, whose history stretches back more than 150 years. Doyle shared an anecdote that captures the company's long connection to communications: GN was the first company to create telegraphic communication between China and the rest of the world. "We like to think that we're still in the world of communicating and connecting people even now," he said.

    Doyle explained that the wider group operates across three areas: hearing, gaming through SteelSeries, and enterprise communications, where Jabra's frontline business sits.

    For many enterprise users, Jabra is already familiar through headsets, speakerphones and video collaboration products traditionally associated with desk-based and knowledge workers.

    Five Things Friday

    Five useful signals from the people shaping global retail. Every Friday.

    By subscribing you agree to receive Five Things Friday emails and to our Privacy Policy. Unsubscribe at any time.

    Now, the company sees an opportunity to bring that audio expertise into frontline environments, particularly retail.

    The rationale is not simply that store associates need better headsets.

    Retailers are progressively digitising the frontline.

    Platforms from companies including Microsoft and Zebra Technologies are bringing more communication and workflow capabilities to employees through handheld devices and applications, including push-to-talk functionality through products such as Workcloud Sync and Teams Walkie Talkie.

    Jabra sees itself as the audio layer connecting workers to those systems.

    And as AI agents become part of those workflows, Doyle believes the importance of that layer will increase significantly.

    Voice Could Become Retail AI's Most Natural Interface

    There is a practical reason Jabra is betting on voice.

    Frontline workers rarely spend their shifts sitting in front of keyboards. They walk, pick, pack, replenish, serve customers, check inventory, move stock and respond to colleagues.

    The nature of the job makes conventional computer interfaces inherently restrictive.

    Doyle argues that voice can eliminate some of that friction.

    He cited figures suggesting people can process around 500 to 600 words per minute in thought, speak at around 200 to 300 words per minute, but type at only 40 to 70 words per minute. "We speak four times faster than we type," Doyle said, with the gap widening further when typing on small handheld devices. His conclusion is that voice is likely to become an important interface for AI-powered applications and agents.

    For retail, the implications are significant. Imagine an associate helping a customer who asks whether another size is available. Instead of stopping the interaction, pulling a device from a pocket, unlocking it, navigating an application and entering a search, the employee could potentially ask an AI agent verbally.

    Or consider a worker replenishing inventory. A voice-enabled agent could provide instructions or receive updates without forcing that employee to stop what they are doing.

    Jabra describes the principle as "heads-up and hands-free."

    In a frontline environment built around constant movement and multitasking, that could make AI considerably more practical.

    The AI Can Be Brilliant and Still Fail Because of the Microphone

    The more consequential part of Doyle's argument concerns accuracy.

    For human-to-human communication, poor audio is frustrating.

    For human-to-AI communication, poor audio can become a data problem.

    If an associate mishears an instruction from an AI agent, the task may be completed incorrectly. More importantly, if the AI mishears the associate, inaccurate information could potentially enter a workflow.

    That raises the importance of what Doyle calls audio intelligibility.

    Jabra's role, he argues, is not simply making the AI audible. It is ensuring that communication between the human and the machine is accurate enough to be useful.

    Doyle cited Jabra data suggesting an enterprise-grade Jabra product can achieve transcription accuracy of around 96%, compared with figures as low as 64% for a standard consumer-grade headset in the comparison he discussed.

    Those figures should be understood as claims made by Jabra during the interview rather than an independent industry benchmark, but they illustrate the wider issue Doyle is highlighting.

    Once voice starts triggering real business processes, small errors become much more important.

    A misunderstood conversation is inconvenient. A misunderstood command that updates inventory, initiates a workflow or feeds information into another system could be operationally significant.

    Retail's Noise Problem Becomes an AI Problem

    Stores are difficult audio environments. Music is playing, customers are talking, trolleys are moving, stock is being unpacked, machinery may be operating in back-of-house areas and employees are speaking to one another.

    Traditional noise cancellation often focuses on what the headset wearer hears.

    Doyle explained that Jabra also concentrates on what happens in the opposite direction: removing surrounding noise from the audio being transmitted through the microphone. Using noise-cancelling microphone technology and digital algorithms, the objective is to isolate the employee's voice from the surrounding environment.

    That becomes particularly important when the listener is no longer another person capable of interpreting context.

    AI systems depend on the signal they receive. The cleaner that signal, the greater the likelihood that speech recognition and subsequent workflows operate as intended.

    Jabra says its Perform 75, launched at NRF in January 2025, can deliver 99% microphone noise cancellation. That is Jabra's stated product claim, but it demonstrates how the company is positioning audio performance as infrastructure for voice-driven applications rather than merely a feature for clearer conversations.

    The Last Mile of AI Isn't Software

    Doyle shared an example that captures the challenge.

    Shortly after moving into his frontline role, he met with a UK retailer experimenting with AI agents designed to provide employees with guidance.

    The retailer had built what Doyle described as a strong proof of concept. But its IT lead recognised a potential weakness: if the headset did not work properly, the entire experience could fall over. All the technological heavy lifting happening behind the scenes ultimately depended on the quality of the endpoint through which the associate interacted with the system.

    It is an important lesson for the wider retail AI market.

    The industry is understandably concentrating on models, agents, integrations and data.

    But frontline AI has a physical last mile.

    An employee needs to interact with it somehow. If that interaction requires workers to repeatedly stop, retrieve a device and type, the productivity promise weakens. If voice recognition struggles because the environment is too noisy, automation becomes unreliable. If the headset is uncomfortable after several hours, employees may simply stop wearing it.

    The user interface is therefore not separate from the AI deployment. It is part of the deployment.

    Jabra Perform 75 Is Designed Around the Reality of a Retail Shift

    Jabra's answer to that problem is the Perform 75, joining its existing Perform portfolio.

    Doyle said the device has been designed for environments spanning back-of-store and warehouse operations through to customer-facing areas.

    But some of its most interesting features have little to do with AI itself. They address the mundane realities that determine whether enterprise hardware survives on the frontline.

    Battery life is one. Jabra says the Perform 75 delivers up to 30 hours of battery life, enough to cover multiple shifts under expected usage. The product also uses a modular design with a hot-swappable battery. Employees can have their own headband or alternative neckband while shared electronic components can be collected from a charging area when a shift begins.

    Tap-to-pair Bluetooth connectivity is designed to reduce the time required to get employees connected and working.

    These may sound like relatively small details. At retail scale, they are not.

    A device deployed to thousands of workers needs to deal with charging, hygiene, sharing, shift changes, comfort, pairing and replacement. A technically impressive product that creates operational headaches for store managers will struggle to deliver its promised productivity gains.

    Doyle said Jabra deliberately considered shift work, shared-device models, power management and ease of use when developing the product.

    Interoperability Matters More Than Another Closed Ecosystem

    Jabra is also making a deliberate decision about what it does not want to be.

    The company is not trying to replace the handheld devices retailers already use. Instead, its hardware is designed to connect with them.

    Doyle described interoperability as a central requirement, pointing to Jabra's work with technology and push-to-talk ecosystem partners to make its Perform products integrate with existing frontline platforms.

    That strategy reflects the reality of enterprise retail technology.

    Large retailers already operate complicated estates containing hardware and software from multiple vendors. Expecting them to rip everything out simply to deploy a new audio solution would create another barrier.

    The more useful role for Jabra may therefore be as a bridge. The handheld device provides computing and connectivity. The application provides the workflow. AI provides intelligence. The headset provides the human interface.

    If that model develops as Doyle expects, audio hardware becomes less of a peripheral and more of a component in the frontline technology stack.

    AI Could Change What Frontline Communication Actually Means

    Until recently, a retail headset primarily connected one employee with another, directing help to checkout, checking the stockroom, sending somebody to aisle five or calling a supervisor.

    That human-to-human communication will remain important.

    But Doyle believes frontline communication is expanding into human-to-machine communication.

    That changes the potential purpose of the device. The headset could connect an employee to an AI-powered knowledge base, a training assistant, a product expert, a task-management agent, an inventory system, a translation service, a customer-service tool or an intelligent workflow capable of acting on what the employee says.

    Doyle believes Jabra can become a "critical gateway" to AI agents as these systems proliferate across frontline environments.

    That is the bigger bet behind Jabra's move into frontline retail. It is not simply betting that retailers will buy more headsets. It is betting that voice will become infrastructure for frontline AI.

    Better Technology Still Has to Work for the Worker

    There is another important dimension to the conversation.

    Frontline technology often fails when businesses design around the system rather than the person expected to use it.

    Retail employees may wear communication equipment for hours. Devices can become uncomfortable. Poorly designed hardware can interfere with movement. Battery failures create frustration. Connectivity problems destroy trust. And every additional action required to operate a device chips away at the productivity gain it was supposed to create.

    Doyle emphasised comfort alongside audio performance because these are personal devices that retailers may increasingly ask associates to wear throughout their shifts.

    That is particularly relevant as AI reaches the frontline. Successful adoption will not come simply because the technology exists. Workers need to find it useful, fast, comfortable, reliable and easier than the process it replaces.

    The Frontline Could Be Where Voice AI Proves Its Value

    Much of the current AI conversation is happening behind a screen.

    Retail's frontline presents a different test. Employees are mobile, their hands are occupied, their environments are noisy, customer interactions happen in real time, information needs to be retrieved quickly and technology has to work without becoming another task.

    Those constraints may actually make the frontline one of the strongest arguments for voice-based AI.

    The productivity opportunity is not giving employees another application to check. It is giving them access to information and intelligence without forcing them to stop working to access it.

    That is where Jabra believes its opportunity lies.

    The most sophisticated AI agent in the world has limited value if the employee cannot communicate with it naturally.

    And as retailers race to build increasingly intelligent stores, the technology that matters most may not always be the technology attracting the biggest headlines.

    Sometimes it is the microphone sitting centimetres from the worker's mouth.

    Because the future of frontline AI will depend not only on whether machines become intelligent enough to answer.

    It will depend on whether, in the middle of a busy retail floor, they can actually hear the question.