Manifest Vegas: Q&A with FourKites' Sree Mangalampalli

The supply chain industry is currently navigating a pivot point where visibility is no longer enough. As global networks face unprecedented volatility, the focus has shifted toward active orchestration – using real-time data to not just see a problem, but to solve it autonomously.

While many firms are still experimenting with isolated AI pilots, FourKites is doubling down on a networked approach.

By deploying a suite of specialised AI agents, they are connecting the dots between suppliers, carriers, yards and customers, transforming what was once a series of siloed problems into a unified, automated workflow.

We sat down with Sree Mangalampalli, Vice President for Digital Transformation Solutions at FourKites, at Manifest Vegas to discuss the trailblazing use of AI agents, the data scale required to train them – and why he was personally proven wrong about the speed of AI adoption in the supply chain.

Could you tell us about your background and what led you to your current role at FourKites?

I've been in the industry for about 20 plus years. If you take the first half of my career, it was mostly in manufacturing and lean manufacturing. In the second half, I moved into consulting, specifically supply chain transformation and the company I used to work for then moved into digital twins and digital transformation.

When FourKites started foraying into end-to-end supply chain orchestration or the control tower space, they were looking for someone who could come in as a subject matter expert who could actually evangelise AI and be a catalyst for AI. That’s how I got into this particular role.

How is FourKites positioning itself differently compared to other AI players in this space?

FourKites was the first supply chain company to release two AI agents that were already used by existing enterprises such as Unilever and Arrow Electronics; that actually was how we were kind of the first in the space.

We were trailblazers and what we are doing is very different from everything else that you hear around here because we have AI agents that are trained on millions of data points on a daily basis.

FourKites actually tracks and orchestrates about three million shipments every day. So that means the agents that we are using are far better trained and have much more data that they use to orchestrate that end-to-end supply chain.

Can you walk us through the specific agents you’ve deployed and how they interact with each other?

There’s a lot of companies talking about individual agents that are solving a piece of the puzzle.

We actually have agents that are tied from start to finish. Meaning we have agents that help collaborate with the suppliers. Then, when the shipment is in transit, we have an agent called Tracy that will actually collaborate with carriers.

Then we have an agent called Allan that will actually schedule shipments into your yards. We have an agent called Folly that will help with gathering proof of delivery documents. And then we have an agent called Cassie that'll actually help with answering customer-related inquiries. So from suppliers to customers, we have agents that actually orchestrate end-to-end.

Looking ahead, how quickly do you expect AI to handle the majority of supply chain processes without human intervention?

I don't know if you guys have heard the saying from Bill Gates. I think he said companies almost always overestimate what they can achieve in one year and underestimate what they can achieve in 10 years.

I was in a class full of supply chain students a few years back – three years ago – and they asked me, "Hey, how far can AI really help in supply chain?" I actually quoted this and I said that supply chain is too complex; it'll take a while before AI can actually make its way into supply chain. And I was proven wrong.

Within two years, we have come up with these AI agents that are changing the world as we speak. So I would maybe give it two to three more years, and for the most complex use cases maybe five years, but things are moving at a very rapid pace right now.

I wouldn't be surprised if these AI agents can automate a lot of these processes without human intervention. They're already doing that in some cases, but I'm talking about the majority of the supply chain process.