How COOs are finding clarity in 2026 amid ‘parallel universes’ of AI adoption

By 2030, half of cross-functional solutions used in managing supply chains will leverage agentic AI capabilities, one top-tier analyst firm says. It adds that by 2028, AI agents may autonomously make up to 15% of daily work decisions. One third-party logistics firm is leveraging AI to optimize delivery routes in real time, saving over millions of gallons of fuel annually, and a major consumer products company says it has integrated AI across 20 global control towers to improve demand responsiveness.

Stats such as these illustrate how quickly the technology is moving forward and how transformative it can be — as well as create a feeling of unease for many executives. Many executives are barely scratching the surface on how to use agents and physical AI — two large sources of venture capital dollars — to transform supply chains and value chains. Citing a prior conversation with Fortune 250 CFOs, Valente noted that “Everyone feels behind, but they’re in the same place, with some level of movement and urgency. And use cases are waning as a method to actually advance value in the organization.”

When polled on their initiatives, roundtable participants were about evenly divided between early-stage pilots, specific use cases and a holistic approach centered on the supply chain and operations function. Just 10% said their AI efforts encompassed a reimagination of the entire enterprise.

“I’m still struggling to get my hands around all that we could be doing,” one COO said. “We have a specific issue where someone will come forward with a specific tool to solve a specific task. Some are anything from process optimization to predictive maintenance and inventory management. What we haven’t done is say: ‘What’s the long-term goal of how we want to play, and what’s the strategy for moving outside of individual value cases?’ I can’t see it yet for getting to A to B.”

Executives should approach AI through a lens of reimagining processes rather than merely improving them, urged Don Frieson, formerly of Lowe’s and now COO Advisor in Residence at EY CEL. COOs understood the power behind the thought but are struggling to make it real. “I don’t think we’ve matured yet on how to redesign everything,” one said. “I’m on the board of a company that is already there. In terms of their approach, they say: ‘Rather than redesign our customer service platform, how do we listen to our customers and take the top 80% of questions and design a system around it?’ But we’re not there yet.”

Instead, attendees spoke of grassroots efforts to familiarize their workers with tools to solve point problems, such as using agents to evaluate return material authorizations. And the evolutions are happening with the added challenge of maintaining customer-facing operations. “Some of this may be throwaway work as you drive a holistic transformation, but we hope people can experiment and get comfortable with it,” one executive said.

Another leader said he considered AI in four main buckets: general tools that everyone in the organization can use, agents within your major apps, point solutions that could become enterprise solutions, and then bespoke tools for whatever isn’t available. Participants cited AI use cases such as:

• Cameras to identify waste and unsafe conditions on shop floors
• Advanced production scheduling, getting away from manual, spreadsheet-enabled plans
• Inventory optimization, still with a human in the loop
• Risk management on tariffs and trade

To gain traction with transformation, leaders believed that it was best to begin with reimagining a function like procurement, in which AI-driven negotiation on payment terms can occur autonomously. It also helps to orient around an immediate goal like “How do you reduce your contract labor through AI?”

Yet our executives continue to duel with a familiar foe that both enables and thwarts their AI aspirations: data.

“The frustrating thing is that we’ve been working on data heavily for a long time,” one COO said. “But when you get beyond just using the tools and redesigning processes, you find that you constantly have data gaps. You dig this tunnel, and you think it’s straight through. But it’s constantly: ‘We have to get this data.’”

Farooque Munshi, EY Americas Data and AI Advanced Manufacturing Leader, knows these problems intimately. “Your data is locked in a variety of systems, and there’s a lack of maturity on common definitions,” he said. “It’s a learning process on getting the capability out in a way that is more usable and not just a onetime activity for an immediate use case. You have to move to this idea of data as a product.”

Data and trust are critical building blocks for scaling point use cases into end-to-end capabilities — moving from generative, to agentic, to autonomous. Most of our COO participants aspired to having data as a product, with trust built in, but they were overwhelmingly focused on creating a single enterprise data foundation backed with strong governance and ownership.

For Munshi, the three core foundational components are:

• Data as a strategic capability, with seamless accessibility across the enterprise
• Strategy as a north star that encourages modularity
• Partnership between IT and the business about who owns these data assets

“The data governance and structure is the biggest hurdle,” a supply chain leader of a utility said. “We took some advice from a major AI leader about trying out some of these tools in a structured way so people get used to them while we take the time to get our policies and data security set up well. They cautioned us not to let anyone pressure us into doing something we’re not ready for. We want people to build agents, but we have to get that policy and governance set up first.”

These discussions are taking place against a backdrop of fear about AI. “AI is here — it’s intimidating,” one COO said. “People are not engaging the tools with the progress we would want. We’re doing a lot of things to make this more organic, having them experiment, but there’s resistance.”

It’s important to zoom out and remind ourselves that AI is not just an internal debate about operations: it has profound cultural connotations as well across your entire ecosystem. “I think of it in terms of business outcomes for our employees, our customers and our suppliers,” he said. “From an employee perspective, how do employees accept AI? Getting them to understand and believe in what they’ve been giving is a challenge. On the customer side, we’re applying it to pricing — how do customers understand what we’re doing, without surprises? And for suppliers, as we partner together, we want to see those benefits they’re getting from AI shared forward to us. How do we make sure that it adds value for every part of our business?”

That is the core question, and EY leaders developed five convictions shaping AI in supply chains to follow:

1. Build the foundation first. Successful AI adoption demands a strong, scalable tech foundation — covering infrastructure, data and security. AI-ready environments need to be flexible and able to integrate with backbone ERPs and bespoke applications.

2. Shift from “nice-to-have” to “essential.” The age of endless AI experimentation is over — it’s time to deliver tangible results from AI investments. Prototypes must transition to production and tied back to outcomes, KPIs and measurable impact on revenue.

3. Embed AI in supply chain — it’s not a bolt-on. This is particularly important when considering governance, which is vital for agentic AI adoption.

4. Drive AI fluency. Internal education around supply chain AI opportunities and risks is lacking. A task force must make bold decisions and challenge every opportunity.

5. Develop the new digital operating models. Reimagine workflows with AI, empower individuals with AI agents and elevate innovation with human/AI teams — none of which can happen under business-as-usual as we understand it today.