The Rise of Supply Chains That Adapt

When you click “buy now” on online stores, recommendations appear, warehouses spring into action, and packages begin their journey to your door. Behind that seamless experience is a fundamental transformation. Supply chains, once governed by fixed rules and human planning, are becoming AI-driven systems that learn, adapt, and increasingly act on their own.

In her book chapter, Reshaping Supply Chains Through AI-Empowered Automation, Jeannette Song, the R.David Thomas Professor at Duke University’s Fuqua School of Business argues that AI is fundamentally changing how supply chains work.

“AI is reshaping supply chains in four connected ways: expanding automation across the chain, changing how humans and machines work together,” she said, "raising new questions about privacy and accountability, and pointing toward a future of more autonomous, agentic systems.”

From fixed rules to adaptive intelligence systems

For decades, automation in supply chains followed a simple logic: “if X happens, do Y.” For example: reorder when inventory falls below a target, send an order through a preset workflow, or stop a machine when a known error appears. These rule-based systems can be fast and reliable, but not very flexible when conditions shift, Song said.

That model is changing. Today’s AI systems go beyond instructions—they interpret data, update forecasts, and revise decisions as new information arrives.  Song compares this shift to the evolution of navigation apps.

“A static map gives you one route and assumes the world will cooperate,” she said. “A live navigation system keeps checking traffic, accidents, and road closures, then recalculates along the way. Supply-chain AI works more like the second model: it helps companies adjust as demand shifts, congestion builds, weather changes, or a supplier runs into trouble.”

The result is what Song describes as a move from “fixed tasks to adaptive intelligence,” where systems both decide and execute.

Long before the current wave of generative AI, Song’s research showed that inventory policies should adapt when demand shifts across environments, rather than rely on a single static forecast. More recent work with Fuqua’s Bora Keskin extended that logic, developing AI algorithms for adaptive pricing and replenishment decisions as conditions change, adding the ability to detect those shifts continuously from live data.

AI in action

In earlier research, Song showed how operations should respond to changing demand. “AI,” she said, “makes that responsiveness richer and faster.” 

At the consumer level, recommendation engines learn what you browse and buy. Chatbots personalize shopping, answer routine questions, and make returns or delivery coordination easier. These systems shape the customer experience while generating useful demand signals, Song said. 

Retailers are also using AI to move faster from demand signals to product development. Song points to Walmart’s Trend-to-Product system as an example of how generative AI and visual analytics can help detect emerging fashion trends and turn them into viable product concepts more quickly.

In warehouses, companies like Amazon have replaced static storage with mobile shelves guided by AI-coordinated robotics. Workers no longer search for products; “the shelves come to them,” Song said.

On factory floors, companies including BYD, Foxconn, and BMW are combining robotics with AI for tasks such as welding, assembly, and quality inspection.

In sum, these examples illustrate that AI is expanding across every node of the supply chain—from demand sensing and product development to warehousing, manufacturing, and logistics. 

Humans are not disappearing

Despite fears of automation, Song sees a future defined by human-machine collaboration, not replacement.

“Human roles are shifting away from repetitive execution and toward supervision, exception handling, and judgment,” she said.

In high-stakes environments—like air traffic control—human decision-making capabilities remain essential.

“It’s like an airport control tower: a panoramic view of operations, with different teams tracking different signals and coordinating in real time to keep everything running smoothly,” Song said. "As AI absorbs more routine analysis and coordination, managers are more likely to investigate anomalies, and decide when a recommendation should be accepted, modified, or overruled."

The risks beneath the promise

However, as supply chains become more autonomous, new risks emerge.

One is privacy and data governance. AI often depends on large amounts of information, moving across suppliers, warehouses, carriers, and platforms. That raises fundamental questions about who owns the data, who can see it, and how it can be shared, Song points out.

Another issue is explainability. Some AI models operate as black boxes, making it difficult for people to understand how AI recommendations were reached.

“That becomes a serious problem when companies need to audit decisions or intervene quickly,” Song said.

Recent research by Song and Keskin on the use of blockchain for perishable products raised similar questions about who gets access to sensitive information and how automated rules should be enforced.

The next frontier: Agentic AI

Looking ahead, agentic AI is likely to define the next phase of supply chains, Song said.

Unlike traditional systems, agentic AI consists of autonomous, goal-directed agents that coordinate across networks. These systems can sense conditions, make decisions, and execute actions with minimal human intervention.

In supply chains, this could mean networks that dynamically adjust inventory, reroute shipments, or reconfigure production in response to disruptions.

Song’s research on multi-sourcing shows how firms can dynamically shift orders across suppliers in response to uncertainty—a logic that future AI-enabled systems can execute continuously and at scale.

“This future is not magic,” Song said. “It is the next step in a longer progression from static rules to optimized decision models, to systems that can update and execute those models more continuously as new information arrives.”

Who is accountable?

AI is reshaping supply chains by changing how decisions are made. What were once linear, rule-based systems are becoming dynamic, learning networks—capable of sensing, adapting, and acting.

But greater autonomy brings new challenges, especially around accountability. As decisions move from humans to algorithms, firms will have to decide how much autonomy to grant, where humans must remain in the loop, and how to make automated systems auditable and aligned with broader organizational and social goals.

As Song writes, supply-chain systems should be designed to “learn, adapt, and align with broader human and societal values.”