AI in Multi-Site Operations Is Still on the Periphery and the Biggest Barrier Is People

AI adoption in facilities and distribution operations is not starting where most people expect. The early wins show up in order management, collections, accounts payable, and route optimization, not in the physical coordination of warehouses, sites, and field teams. That is not a shortcoming. It is a sequencing problem. The transactional automation happening now is generating the data that cross-functional orchestration will need to work.

Ladji Kouyate is a technology sales leader and co-founder of WISMO, an AI-powered email agent built for small manufacturers that automatically intercepts and answers customer emails. He previously served as General Manager of a $200 million P&L protein business at Cargill, overseeing sales, marketing, finance, and supply chain across a 40-person team. He holds an MBA from Tuck at Dartmouth and an MPA from Harvard's Kennedy School.

"Right now, most AI in operations is still around order management, collecting funds, accounts payable, and accounts receivable. It's around the periphery of operations, not so much the actual moving or managing of locations and how work gets done," says Kouyate.

Siloed today, orchestrated tomorrow

Kouyate describes current AI adoption in operations as early and fragmented by design. Projects in accounts receivable, order management, and warehouse routing run independently, each solving a contained problem. But he sees them converging.

"As we move toward the short term, some of these projects are going to start to be multifunctional," Kouyate says. He describes a scenario where accounts receivable connects to order management so that a system can flag a customer who is likely to reorder based on purchase history, then prompt outreach to settle outstanding balances before the next order. "That is more of the orchestration, but we're really not there yet."

Kouyate emphasizes that data quality today determines what AI can do tomorrow. "A lot of the work being done right now is how comprehensive sales teams are when it comes to capturing data," he says. "We may not be ready to make the AI decision at the moment. But we know we're going to need that data in a year or two when the tools are mature."

The regional layer question

As AI lowers transaction costs, Kouyate raises a structural question that multi-site operators will have to answer: does the regional layer still need to exist?

At a company like Cargill, centralized operations handle large national accounts while regional distribution houses serve smaller customers. "In the past, you would need people in the region to chase a $50,000 order," Kouyate says. "Corporate people chase million-dollar orders. But now, because the cost of transacting is getting smaller, do you get rid of the people in the region and say we can all do it at a corporate level?" The answer is not straightforward, because regional knowledge carries value that data systems do not yet replicate.

Kouyate points to the Sysco acquisition of Restaurant Depot as a case study. Restaurant Depot runs on $300 cash-and-carry transactions with local restaurant owners. Sysco runs on much larger check sizes. "AI and the cost of processing going down should allow you to play in both," Kouyate says. "But if you're not leveraging AI, it is very difficult to go from a $100,000 average check to $300." The technology makes the economics viable, but operating behaviors have to change with it.

People are the constraint

When asked about the biggest barriers to scaling AI-driven operations across multi-site environments, Kouyate does not point to technology. He points to people.

Regional operators make decisions based on market reputation that centralized systems cannot quantify. Kouyate gives an example: a regional general manager might take a customer to small claims court over $300 to send a signal. "All of these restaurant owners know each other. They're all families," he says. That kind of contextual judgment does not transfer into an AI model.

The Sysco and Restaurant Depot deal illustrates the tension. "The cultures at play are going to be very interesting," Kouyate says. Restaurant Depot operated with a scrappy, cost-cutting mentality. Sysco operates with higher quality standards. "If you look at those margins and think you're going to capture them all, you're 100% wrong," he says. "As you try to do the right thing, those margins are going to narrow."

On the hardware side, Kouyate notes that the cost of tracking physical networks is falling fast, with fleet tracking and distribution monitoring devices becoming cheap enough for broad deployment. But he is clear that the distribution is still in pilot mode.

"A lot of what we're looking at are pilots right now," he says. "Manufacturing is different; those are million-dollar investments. Distribution is a different beast." The technology is moving toward readiness. Whether the organizations behind it can adapt fast enough to use it is the open question.