Enterprise investment in artificial intelligence (AI) is accelerating, but many organizations are still struggling to move beyond scattered pilots and into organization-wide adoption.
The gap between ambition and execution often comes down to workforce readiness rather than the technology itself. Deep AI adoption is a multi-year undertaking, not a quick rollout, and change management often lags behind deployment.
“AI bloat” refers to over-engineering, inefficiency or clutter caused by integrating AI without a clear strategy. It echoes the earlier problem of SaaS bloat, where organizations accumulated overlapping software subscriptions and rarely used features, making systems harder to manage.
“How do we avoid ‘AI bloat’ the way we saw SaaS bloat?” Matt Ausman, chief information officer of Zebra Technologies, told TechJournal.uk in an email interview. “What is a realistic workforce adoption goal for year one versus year two? How long does change management truly take?”
“We started 2026 expecting over 80% of the organizations to be using AI tools at least once a month,” he said. “By the end of the year, the bar will rise significantly, with 80% using it weekly and over 50% daily.”
He said ambition is currently outpacing capability across most organizations, with only about 5% of staff typically working at the leading edge of new technology.
“There’s the bleeding edge, the leading edge, or you may make a conscious choice to lag on the technology front. I’ve chosen the leading, not bleeding, edge,” he said.
Zebra has deliberately turned down many newly released tools, he said, relying instead on standard tools already deployed broadly across the business to learn what is achievable before scaling further.
Ausman outlined a five-step blueprint for enterprise AI adoption:
Create a cross-functional governance council spanning HR, legal, finance and business units
Appoint a full-time executive to lead AI transformation
Define a clear strategy for technology adoption and cost accountability
Invest in training for both leaders and frontline employees
Launch a holistic change management program with clear communication and feedback loops
“What is important is to choose your strategy deliberately and stick with it,” he said.
Zebra Technologies is a global leader in frontline and industrial innovation, providing connected devices, software and automation solutions across sectors including retail, manufacturing, and transport and logistics. Ausman leads the company’s global IT strategy and its internal use of generative AI (GenAI).
Frontline AI gathers pace
Zebra’s global research with Oxford Economics examined how AI is being deployed across frontline industries including retail, manufacturing and logistics, tracking adoption from early piloting through to full-scale use.
Retailers showed the highest levels of AI deployment and piloting across inventory management, cost control, demand forecasting and personalized offers for shoppers, according to the research.
“Retail is an industry that moves fast, with high direct consumer engagement and B2B across supply chains,” Ausman said.
He said retail customers are using AI to create personalized offers on shopper self-scanners, while others deploy on-device agents to support inventory, merchandising and staff training.
Manufacturers recorded somewhat lower adoption than retail and logistics, though with similar levels of piloting activity, reflecting more closed and tightly regulated working environments.
Predictive maintenance, product quality intelligence, and inventory optimization ranked highest among manufacturing workflows, Ausman said, alongside the growing use of machine vision for complex visual inspection tasks in textiles and food processing.
In logistics, nearly 40% of respondents use AI tools for demand forecasting, and over two-thirds deploy or pilot the technology for inventory management, the research found.
“Nearly two-thirds are deploying or using the technology for predictive estimated time of arrival, while 57% are doing so for route planning and optimization,” Ausman said.
Beyond individual workflows, Ausman pointed to a broader shift he calls ambient intelligence (AmI).
AmI is a computing model in which AI and sensors are invisibly embedded into physical environments. These responsive ecosystems, such as smart homes, hospitals and offices, passively recognize human presence and adapt to individual needs without manual prompts or app interactions.
“If the AI Internet-of-Things (IoT) is the ‘how,’ then ambient intelligence is the ‘what,’ the valuable outcome that frontline workers can leverage in their jobs,” he said.
Processing this data at the edge with on-device AI models allows for more autonomous operations and near-instant workflows, he said, particularly in industrial and outdoor settings where connectivity can be unreliable.
“This combination of human worker experience and judgment, AI models and AI agents can be described as augmented collective intelligence,” Ausman said.
But Ausman was candid about the human cost of this shift. As AI frees employees to focus on more strategic and creative work, he said, organizations must stay alert to cognitive endurance and the risk of burnout.
“This is a potential ‘pyrrhic victory,’ where efficiency gains are offset by a decline in well-being,” he said. “We must design roles that balance this cognitive load to ensure a sustainable and engaged workforce.”
From pilots to practice
Ausman pointed to two distinct ways Zebra is putting GenAI to work internally, one that augments existing jobs and another that rebuilds processes from the ground up.
“Our Frontline AI Companion for warehouse workers is an example,” he said. “We’ve created agents that answer questions, provide recommendations and automate specific tasks.”
Zebra recently deployed a new agent that determines the optimal way to pack products onto a pallet, he said, representing the first of the two approaches by augmenting a single human work step rather than replacing it outright.
“Old, multi-step process for scanning inbound pallets has been completely reimagined,” he said. “Now, using a Zebra device, an employee can simply flip through shipping documents with the camera, and AI recognizes the different pages.”
“They then hold the camera up to the pallet, and the AI determines barcodes, box labels, and quantities,” he said. “If more information is needed, the AI will prompt the employee to walk around the pallet.”
The enterprise resource planning (ERP) system updates automatically once the scan is complete, he said, removing manual data entry from the process entirely.
“It’s a model we can apply to customer service inquiries, IT support ticket triage, and even marketing content creation,” he said. “In the near future, AI agents will be the new email; we’ll wonder how we ever got work done without them.”
Zebra has also launched an AI zPartner Assistant for its PartnerConnect partners, he said, helping them build marketing and business plans, customer proposals and account activity reports.
“This is shifting AI from a simple tool to a more symbiotic relationship where a human and a digital solution work collaboratively together,” he said.
The role of the CIO itself is changing under the weight of this shift, Ausman said, moving well beyond a purely technical remit.
“A CIO today cannot simply be a technologist,” he said. “We must be a pathfinder, tracing a new path in an uncharted field, guided by experience from past technology shifts like the internet and the cloud.”
“We must also be an evangelist and a C-suite advocate, championing the vision and securing the buy-in necessary for true transformation,” he said. “My approach is simple: centralize governance, federate innovation, and measure value.”
The CIO and IT organization increasingly need an equal seat at the table with broader business teams, a shift that has been building for years but has accelerated with AI.
On June 8, the company expanded its software portfolio by launching Zebra Nucleus, a unified platform for managing device fleets, alongside new Workcloud tools that provide frontline teams with real-time, AI-powered insights and orchestrated workflows.
Zebra’s own research suggests the industry is still early in this shift, with organizations across retail, manufacturing and logistics expected to move from scattered pilots toward broader, AI-driven operations over the next several years.
Solid analysis on the gap between procuring AI tools and driving actual operational ROI.
A critical risk when addressing enterprise AI bloat:
- The Pendulum Risk: Swinging too far toward central IT control doesn't eliminate bloat; it just drives it underground into Shadow AI.
- The Delivery Bottleneck: Heavy gatekeeping forces frustrated teams to rely on unsanctioned personal tools to maintain execution speed.
The Pragmatic Fix: Move from heavy front-end gatekeeping to active engagement-based pruning. Automatically offboard any pilot that fails to hit defined usage benchmarks within 60 days.
How is Zebra currently structuring the balance between pilot speed and firm offboarding criteria?