Every CEO is talking about AI right now. I get it, I do the same thing. But most of that conversation skips the part that matters most in supply chain: knowing exactly where AI stops being useful and where you still have to show up yourself.
Where I See AI Earning Its Keep
I'm not an AI skeptic. I use it daily, and if you're a CEO, a COO, or a CFO and you're not using a Claude or an OpenAI for something right now, I think you're missing out. But I'd separate the real value into two areas, and it's worth being precise about both.
The first is rapid insight, the ability to look through large volumes of data quickly and get something useful back. This is genuinely powerful, but it isn't infallible. I run deeper financial analysis through Claude regularly, working across information I have access to, including financials for our business, and often I get an answer back that's wrong. I catch it because I understand the context of what I asked and how the data works underneath it. If you didn't have that grounding, you might look at that wrong answer and think it looked right, and then make real decisions off of it.
The second is operational execution, which I think of as the next evolution of robotic process automation. You don't need to map out every step of a process anymore. You explain generally how it's supposed to work, define the decision points, and keep training the model while it handles repeat operations. I've been interacting with more support agents in my personal life lately, and the communication is excellent, instant, and available 24 hours a day.
Insight and execution, that's where I'd tell any executive in supply chain to focus right now. Here's the part that comes next.
The Piece AI Wasn't Built For
One of our top account executives here said something to me recently that stuck. She was talking through a major account, and she said, "The one thing that AI can't do is accountability."
She's right, and the reason comes down to how supply chains are actually built. Multiple vendors, multiple modes, multiple carriers, multiple suppliers, all moving at once. It is not designed for perfect execution, which means the relationship and the accountability still have to be there, held by people.
This is why we still run program services for some of our customers. I'm upfront with them about it: you can run everything yourself in the platform, fully self-service, and it'll work. The bigger reason you'd want us involved is that Trax is managing more than $20 billion in transportation spend. That gives us a lot more ability to get a carrier to do something, or to see past the one small problem in front of you and address a broader problem that carrier may be having. I can help our customers be more accountable to themselves as well.
AI gives you great alerting, early warnings, and patterns worth flagging. What it can't do is sit across the table from a carrier and hold them to a commitment. I don't know when that changes, if it ever does. Many years from now, maybe. Or maybe not.
What I'd Tell You To Do With This
My advice for separating real AI opportunity from hype starts by setting the technology itself aside and starting with the business problem. In logistics, you're always working with four levers: cost, quality, speed, and emissions. Pick the one that matters most to you right now, and go after a specific suspicion you already have about where you're losing ground. Good news, you can use AI to find out if that suspicion is right or wrong.
What doesn't work is a blanket strategy. If somebody tells me they're going to "AI everything," I think your alarm should go off, especially in supply chain. I want that ambition. I just need to see the use cases that will pull themselves out of the hype and solve a real problem.
That distinction, insight versus ownership, is what I think will separate the companies that get real value out of AI from the ones that just get a faster way to be wrong. AI can hand you the insight all day long. Ownership is still on us.
Take This Further
Free up your team to do the higher-level work AI can't touch, but know exactly what you still have to own yourself. Watch the full conversation on the Trax YouTube channel, or find more of my thinking on this on the Trax blog.
