Inside Zebra Technologies’ approach to frontline AI

• Zebra Technologies says labour and supply chain pressures push AI to the frontline.
• Visibility and AI reshape frontline operations in retail, manufacturing, and logistics.

Frontline work is changing, not because companies want it to, but because many not have a choice. Labour is harder to find, customer demands are rising, and supply chains remain hard to predict. At the Zebra Technologies Sales Kick-Off APAC 2026 in Incheon, Korea, the company laid out how it sees these pressures shaping the next phase of frontline operations, and where AI is starting to fit into day-to-day work.

Tom Bianculli, Zebra Technologies’ chief technology officer, framed the discussion around a simple idea: productivity happens where work actually takes place. That includes shop floors, store aisles, warehouses, loading docks, and delivery routes. Zebra has spent decades building tools for those environments, starting with barcode scanning and on-demand thermal printing, which still underpin many track-and-trace systems today. Over time, that foundation has expanded to include rugged mobile computers, robotics, RFID, and machine vision.

RFID, in particular, is seeing faster uptake as companies look for more continuous ways to track goods. Bianculli shared a data point that highlights the pace of change: around 20 billion RFID tags were shipped globally in 2020. By 2028, that number is expected to pass 110 billion, with roughly 80 billion already shipped this year. As more products and assets are tagged, companies are moving beyond one-time scans toward ongoing visibility in supply chains.

Retail remains Zebra’s largest market, accounting for about 30% of revenue, followed by manufacturing and transportation and logistics. Manufacturing has grown quickly over the past year, with Asia-Pacific – especially Korea and India – playing a key role.

Turning visibility into action

Across these sectors, Bianculli said customer concerns tend to cluster around four themes: labour availability and cost, changing customer expectations, supply chain unpredictability, and the need to protect margins by improving productivity. Rather than treating these as separate problems, Zebra groups them under what it calls intelligent operations.

“From an intelligent operations perspective – and to clarify, we didn’t invent the phrase – what it really means is bringing AI, data, and human expertise together to optimise workflows,” Bianculli said.

In manufacturing, that approach shows up clearly in quality inspection. Machine vision cameras mounted along production lines capture images as products move past at speed. AI models analyse those images to spot defects or anomalies. Human expertise is used upfront to define what “good” looks like and what counts as a defect. That knowledge is trained into the system and applied automatically, replacing checks that might otherwise be manual, inconsistent, or skipped altogether.

The same idea carries into warehouses and logistics. Mobile devices are not just tools for scanning barcodes. Newer models act as sensors, capturing spatial and 3D data that can be combined with RFID. “The mobile computer itself becomes a sensor,” Bianculli said, pointing to devices that can collect 3D information, analyse it, and trigger a next action. In practice, that might mean confirming pallet contents, checking asset placement, or helping ensure goods move through inbound and outbound flows as planned.

Retail as an early testing ground

Retail is emerging as one of the earliest areas where these ideas are taking hold. High staff turnover creates pressure to help workers get up to speed quickly and carry out tasks consistently. Stores already rely on mobile devices for task assignment and communication, which makes it easier to layer AI on top.

This is where Zebra Companion comes into play. One part of the system, the Knowledge Assistant, is designed to surface store policies and procedures when workers need them. If an associate is unsure about a return scenario, the assistant can provide guidance for that specific case. Another part, the Sales Assistant, focuses on product information. It allows store staff to get accurate answers while standing with customers in the aisle.

“These were good ideas just a year ago, but now they’re being actively piloted and beginning to scale in multiple stores in real-world environments,” Bianculli said. Zebra is also working on bringing AI-based avatars into self-service kiosks, with the aim of making the experience feel closer to speaking with a human expert.

Scaling AI at Zebra Technologies from pilots to everyday use

Many of these use cases fall under what Bianculli referred to as physical AI. While the term often brings robotics to mind, he said it also includes fixed systems like machine vision cameras and RFID readers that capture data from the physical world. Zebra’s acquisition of Photoneo added high-speed 3D imaging used on robotic arms for tasks like picking, placing, and packaging. Combined with work around spatial data standards, this data helps form digital models of real environments that can support simulation and coordination.

Still, Bianculli was clear that technology alone is not enough. One of the most common barriers to AI adoption is data preparation. His advice was to start small. “Our approach starts with choosing specific use cases,” he said, pointing to an MIT study that found teams were more likely to succeed when they focused on defined workflows and worked with partners not trying to build everything themselves.

“Trying to cleanse and harmonise all enterprise data at once is daunting,” Bianculli said. “When you focus on a specific workflow, you can identify the exact data needed and prepare it appropriately.”

Return on investment often follows the same pattern. Early value tends to come from time saved per task. In delivery routes, for example, AI can help drivers identify the right package faster or automate proof-of-delivery steps. Over time, cost avoidance becomes more important. Verifying shipments at each hand-off can help catch errors early, reducing fines, penalties, or rework later in the process. For many enterprise customers, Bianculli said a 12 to 15 month ROI is a common expectation.

As AI becomes more embedded in frontline work, Zebra expects a shift toward systems that anticipate issues not respond after the fact. Agent-based systems, demand forecasting, and machine learning are starting to guide decisions about what should be where and when. Bianculli described this as a move away from visibility alone, toward coordinated action in people, assets, and systems.

The shift ties into what he called augmented collective intelligence (ACI). Instead of aiming for machines that replace people, the focus is on systems where AI handles automation and coordination, while humans provide judgement and oversight. There is no single all-knowing model. Instead, domain-specific agents support environments like stores, warehouses, and docks, working with people who know those spaces best.

Ryan Goh, who now leads both Zebra’s Asia-Pacific business and its global OEM unit, said these pressures are becoming harder to ignore in the region. “The pace of change in APAC is accelerating rapidly,” he said. “Businesses, especially those in India and Japan, are facing pressure from increasingly disrupted supply chains and rising labour shortfalls.”

Goh added that the challenges customers face are often similar, regardless of geography. What differs is maturity. Some organisations are still largely manual, while others are further along in digitising and automating work. The goal, he said, is to match solutions to where customers are today and help them progress over time.

At the network level, this matters most for large enterprises running hundreds or thousands of sites. Using data from individual locations and applying it consistently in the network allows companies to move beyond isolated improvements. It is this ability to scale insight and action, not any single tool, that Zebra sees as central to the next phase of frontline operations.

As Bianculli summed it up, the work ahead is less about bold promises and more about steady progress – using visibility, AI, and human knowledge together to improve how work gets done, one workflow at a time.

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. This comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.