What Supply Chain Leaders Actually Need to Know About AI

I've spent my career in supply chain, from the factory floor to the C-suite at Clorox and Frito-Lay, and I've seen a lot of technologies come and go, each promising to transform the way we plan and operate. So, when I say that ketteQuest 2026 was one of the most energizing events I've attended in years, I mean it. The sessions were outstanding across the board. But the one that really hit home for me was Chris Amet's session on the Future of AI in Supply Chain Planning.

Chris is ketteQ's CTO, and he's been building planning systems for complex global supply chains for four decades. He didn't come to the stage with a polished pitch deck full of promises. He came with something rarer: hard-won honesty about what AI can actually do today, where it breaks down, and what it means for the people who lead supply chain organizations. As someone who has lived through the ERP wave, the outsourcing wave, and the digital transformation wave, I found his framing both refreshing and, frankly, clarifying.

Here's what I took away.

We Are in a Step-Function Moment, But It Has Limits

Chris opened with a truth that many executives need to hear: AI is genuinely transformative right now, and it also has real, well-defined limits. The last three months, in his words, have brought changes that two full years of AI experimentation did not deliver. Code generation that topped out at 1,000 lines can now reach 30,000 or 40,000. Workflows adapt in real time as supply and demand shift. Analysis that once took skilled planners hours now happens in seconds.

But Chris also named what he calls the 80% rule, and it's the most important framing I've heard on this topic: AI is genuinely good at roughly 80% of supply chain work. And that last 20% is where the most consequential decisions live.

For executives deciding where and how to leverage AI in their organizations, that is the number to hold onto.

Three Failure Modes You Need to Know

Chris was direct about where AI breaks down at the enterprise level, not in theory, but in practice, based on what the ketteQ team has encountered building production-grade systems.

The 10-Iteration Problem. When working through complex planning logic, AI tends to fix one issue and create another. Like squeezing a balloon, every gain somewhere creates pressure somewhere else. The more complex the supply chain, the more pronounced this becomes.

Confident and Wrong. AI doesn't flag what it doesn't know. It fills in blanks. In supply chain planning, where safety stock calculations and inventory parameters carry real operational consequences, a model that fabricates a confident answer isn't a productivity tool. It's a liability. Chris was unequivocal:

"We can't have the AI fill in the blank and say, 'I thought the safety stock was 10, so I just said 10.' That's not acceptable."

The Skeuomorphic Trap. When AI anchors on a particular approach, redirecting it is surprisingly difficult. In complex planning scenarios, this rigidity can cost significant time and erode trust in the output.

The takeaway for leadership: AI deployed without strong guardrails and human oversight introduces risk, not just capability. The organizations that will benefit most are those investing in making AI trustworthy, not just fast.

The Conductor, Not the Orchestra

The analogy that stuck with me most from Chris's session was this: the role AI creates for humans isn't that of a musician. It's that of a conductor. You don't need to know how to play every instrument. You need to know what good sounds like and when something is off.

I've thought about this a lot since ketteQuest. In my experience leading supply chain organizations, the most valuable people were never the ones who knew one thing deeply. They were the ones who could see across the whole system, who understood how a sourcing decision rippled into service levels, or how a demand signal should change an inventory posture. AI amplifies that kind of range. It doesn't replace it.

Chris called the danger "AI psychosis," the initial euphoria of seeing what these tools can do, leading organizations to hand over the baton entirely. I've seen the equivalent in every technology wave I've lived through. The antidote is the same as it's always been: clear accountability, disciplined governance, and a culture that treats AI output as a starting point, not a final answer.

Where Planners Create Irreplaceable Value

Chris mapped the AI opportunity across demand, inventory, and supply, and was precise about where human judgment remains essential.

In demand planning, AI handles pattern recognition and routine forecasting well. Planners own the novel situations: new products, new customers, and the outliers that fall outside any historical pattern.

In inventory planning, AI can continuously run scenarios and surface recommendations. But the planner decides what inventory posture actually fits the business, whether to shift strategy, when to act, and what trade-offs the organization is willing to make.

In supply planning, the most complex domain, AI is a powerful analytical partner across multi-echelon networks, multi-sourcing, and hard constraints. But the decisions carry business accountability that only a human can hold.

The clearest example Chris offered: an allocation decision with limited supply and competing customer demand. AI can model the scenarios. It cannot be accountable for the choice. "As a human, you have to live with that decision. The AI does not." That line has stayed with me.

The Leadership Imperative

Chris closed with a call to action, I echo directly to every supply chain executive reading this: don't wait, get going, keep going.

The organizations that will lead in AI-enabled supply chain planning aren't waiting for the technology to mature. They're learning to use it now, understanding its limits, building governance to make it trustworthy, and developing the human capabilities that AI cannot replace.

One practical insight that surprised me: communicating effectively with AI requires the opposite instinct from what most executives have. We're trained to be concise, to get to the point. With AI, you need to give it more context, not less. Richer prompts. More information. "When you think you've said enough," Chris said, "say more."

The wave is here. The question isn't whether to engage. It's how quickly your organization can build the muscle to do it well. Watching Chris walk through what ketteQ has built and learned, I left that session more convinced than ever that the companies that treat AI as a strategic capability, not just a productivity tool, are the ones that will define what supply chain excellence looks like in the years ahead.

ketteQuest is ketteQ's annual supply chain conference, bringing together planning leaders, technologists, and practitioners to explore the future of intelligent supply chain. The 2026 theme, Agents of Possibility, centered on the practical and strategic implications of AI agents in enterprise planning environments.