How Zebra Technologies Unites Cross-Sector AI Strategies

Achieving true value from AI requires navigating a complex matrix of data hygiene, corporate culture and workflow integration. 

For example, while the retail sector rapidly deploys AI to optimise inventory and elevate customer experiences in the aisles, manufacturing relies on structured, deep-learning machine vision to drive precision on the factory floor. 

Despite these differing environments, a cross-pollination of strategies is occurring, uniting both sectors in a shared goal to build a more connected, real-time workforce.

Here, Zebra Technologies leaders Jason Harvey (Vice President, Automation and AI EMEA), Stuart Hubbard (Global Senior Director, AI and Advanced Development) and Mark Thomson (Retail Strategy Director EMEA) explore how ambient intelligence is elevating human capability, capturing ROI and reshaping the future of operations.

Question 1: From a global AI leader’s perspective, what has been the tipping point for moving AI out of the lab and into the fabric of everyday business?

Stuart: Zebra has had AI in production for many years using machine vision and computer vision. Since the launch of ChatGPT, we’ve seen an acceleration in advances across model research and engineering, the shift from large to small language models and the rise of on-device multimodal AI. 

These developments are coupled with a maturing of the market from consumer app to enterprise-ready solution and the scalability to meet different enterprise requirements. Organisations also needed time to understand the technology, build the business case, evaluate the ROI and put appropriate governance in place. Compared to previous technology supercycles, I think AI adoption has been rapid, and we’re still at the very beginning of the journey.

Mark: Technology often moves faster than the organisations looking to leverage its capabilities. Vision, strategy and executive sponsorship, business process transformation and company culture are vital for AI success, and need time to get right. As does data cleansing, which is an essential element of any AI deployment.  

I would say AI adoption in retail has moved at pace in some areas such as shopper apps, though it is more gradual for employee empowerment so far. Industries like retail are beginning to add AI to current frontline workflows to gain efficiency and productivity advantages. With time, AI will lead to new ways of working and will help to transform those who adopt it, leading to great leaps in efficiency and engagement.

Jason: Organisations re-evaluated their other technologies and now see new value with AI. Data from things like device usage, geolocations, errors and defects, product compliance, inventory movement, task completions and customer engagements are valuable sources to train multimodal AI models. These deliver new types of intelligence and personalisation for frontline workers, customers and operations. AI is being used in combination with RFID systems, machine vision solutions and augmented reality, which adds value to data capture solutions and the volume, variety and velocity of edge data available to AI.   

Question 2: How is this transition manifesting differently in retail aisles versus on the manufacturing floor in terms of long-term strategy?

Mark: Retail is a fast-moving industry with high-volume direct customer engagement in store and online, along with business-to-business engagement across supply chains. AI strategies are being executed, with solutions already widely used and piloted across inventory and cost optimisation, enhanced search and upsell.

Customer experience strategy is also benefitting from AI capabilities, which can positively impact sales. For example, shopper self-scanners are very popular with shoppers, and we’re seeing supermarkets extend value by using AI to analyse shopper behaviour and purchasing data to deliver more personalised recommendations on the device screen as the shopper moves around the shop.

This could extend to dynamic advertisements and offers across retail media networks in store, based on AI analysis of shopper behaviours correlated to days, times and external factors like public holidays and sporting events. This type of strategy can elevate the customer experience and positively impact sales.

Jason: Manufacturing comes with a lot of structure and repeatability, but it’s also a much more closed environment compared to retail. This brings its own set of strategic opportunities and requirements for AI, including getting IT and OT to agree on where to focus AI investments. Our industry research – Elevating manufacturing value: The impact of Intelligent Operations – finds that manufacturers are using AI today for product quality intelligence and inventory optimisation, among other workflows. 

For example, machine vision smart sensors with pre-trained deep learning models on-device can perform optical character recognition and anomaly, and defect detection. This automates a demanding and important job and elevates quality, as deep learning machine vision provides very high levels of accuracy at sub-millimetre levels.

High-quality data is also a foundational issue for AI success in manufacturing. Our research finds that most manufacturers say structured data analysis is performed only in select areas or remains limited and siloed. Some are further ahead: one-fifth say their data management process is automated and performed across multiple functions, and 11% say theirs is fully integrated throughout the organisation with AI insights.

Question 3: We often hear about AI replacing workers, but Zebra’s vision focuses on augmented collective intelligence (ACI). How are your respective industries specifically using AI to support rather than replace frontline teams?

Stuart: To briefly explain, ACI is a conceptual framework I use to explain what Zebra and its partners see and hear when collaborating with customers. There are three key components of ACI.

First, agent swarms or ‘digital workers’ instead of a single ‘all-knowing’ model. ACI utilises a network or ‘swarm’ of connected, dedicated agents.

Second, it is multi-style, combining different styles of AI such as generative and classical algorithms to address complex tasks. And third, human integration, with workers contributing unique intelligence, common sense and domain expertise to the network, while AI scales these talents through decision support and automation.  

Mark: Workers need a culture where they can test and learn with AI, including allowing for a margin of failure. Simply deploying AI tools will not lead to success by itself. The right culture and change management processes will support AI transformation initiatives, but without it, the narrative of fear and low user adoption becomes a risk. Employee understanding and adoption are foundational for grassroots innovation as workers become more confident.

Jason: I’d add to Stuart’s point by focusing on how today’s multimodal AI models are adding new value to current IoT solutions to deliver what we call ‘ambient intelligence’ across retail, logistics and manufacturing. Think of ambient intelligence as digitised physical environments for networks of agents to work within. As mentioned above, organisations are finding new value in their frontline data and data capture solutions.

Voice, video, photos, text, location, movement, inventory, defects and anomalies can be captured using scanners, cameras, machine vision, mobile devices and RFID systems, and ingested by AI models. The models turn data at the edge into intelligence on the frontline to augment human decision-making and to guide agentic workflows. 

Question 4: Where are you seeing the most exciting cross-pollination of AI strategies between retail and manufacturing today?

Stuart: Fundamentally, organisations including retail and manufacturing share similar challenges and goals. They are aiming for more connected frontlines where workforce, customers and data are linked and updated in real-time. Manufacturing plants and retail stores both need high levels of visibility into their assets and inventory – to better manage demand and supply chains, and eliminate waste and loss. 

IT and OT leaders are focusing on moving from traditional, rules-based automation to automation that can self-improve, deliver more insight and requires less human intervention – the AI moves from being initiated by the frontline worker to the AI being a thoughtful partner who anticipates the frontline workers’ needs using data from across the whole environment and supply chain. From these shared challenges and goals, we begin to see how learnings around AI can be shared across different industries.

Mark: This cross-pollination is the value we bring to retail, manufacturing, healthcare and logistics customers in areas like workforce mobility with mobile computing and advanced inventory management transformation with RFID. We are seeing the same now, with AI vision-guided automation moving from manufacturing into retail logistics. AI software can be combined with robotic arms, 3D sensing and smart cameras for picking, sortation and quality inspection use cases. 

Jason: AI application building blocks – ‘enablers’ – can be used by retailers and manufacturers to develop applications integrated onto mobile computers connected to manufacturing execution systems, retail inventory databases and workforce task platforms. A worker can point their device camera at a storage space filled with containers and boxes covered in serial numbers, text and barcodes. 

The AI-powered barcode localiser and text recogniser models see the barcodes and text and identify the correct items to pick, with an augmented reality layer on the screen telling the worker what’s inside the box and additional intelligence like special handling requirements or hazardous contents such as chemicals.  

Question 5: To bring this to life, could you each share a recent customer success story where AI didn’t just improve a metric, but fundamentally changed how that business operates or how its employees feel about their work?

Stuart: The growth and convenience of e-commerce has led to a huge rise in parcel drop-offs to shops, lockers and doorsteps. And it’s here that Zebra’s Proof of Delivery AI Blueprint is making an impact. 

A global logistics leader has given its last-mile delivery drivers Zebra mobile computers fitted with the Proof of Delivery AI Blueprint to improve parcel deliveries to homes. Each driver needs to take a photo as part of the chain of custody and proof of delivery. Sometimes, photos gave rise to customer concerns around privacy. With the AI solution, on-device AI analyses the photo in real time, ensuring it meets quality standards and is free of any sensitive information, like people or house numbers. 

This intelligent automation empowers the driver to complete their task efficiently while delivering a superior customer experience. The result is a dramatic reduction in customer enquiries, improved productivity and a stronger sense of trust in the delivery process.

Mark: Labour hiring, training and retraining are ongoing challenges for retailers. It’s hard to hire enough workers, it takes too long to get them up to speed and there are high levels of churn, including departures to rivals offering a better hourly rate. 

One retailer is addressing these issues with AI. This European supermarket chain hires lots of staff during peak seasons to meet higher shopper footfall and additional seasonal inventory. However, previous onboarding was fragmented and time-consuming to deliver and took time away from consumer-focused tasks. 

With Zebra AI Knowledge Agent running on employee Zebra mobile computers, team leaders were able to provide updated and consistent onboarding and guidance to new hires. The AI agent is trained on the retailer’s standard operating procedures, so new hires can quickly and accurately learn about the company, their role and policies to follow. 

This AI-first approach has delivered important employee experience improvements and faster time-to-value – and team leaders recover time for other activities. It also instils confidence, as frontline staff are in a better position to do their jobs and help customers on an ongoing basis. 

Jason: EBI Electric, a manufacturer in wood production, took advantage of a gap in the market to provide customers with an enhanced offering. It improved supplier relationships and generated over £500,000 (US$670,800) in annual savings. Previously, its legacy technology and outdated operations led to false positives, and increased waste and labour hours spent on manual checks. Its leaders also noticed that competitors weren’t offering upgraded packages to customers. This combination of human insight, operational change and the right culture resulted in a new deep learning and 3D sensing machine vision solution. 

The system plugs into production lines for consistent, reliable results under harsh industrial environments, and the AI continuously learns so precision and yield improve over time. Outcomes include an eightfold improvement in precision, reduced errors, six-figure cost savings and the confidence to expand its business into new global markets.