Reacting to risk: AI’s role in supply chain risk management

From the March-April 2025 edition of Supply Chain Management Review.

March-April 2025

Inside this month's issue of Supply Chain Management Review, we look at the complicated process of managing parts for military aircraft and what private sector supply chain managers can learn. Plus, understanding…

Browse this issue archive.

Access your online digital edition.

Download a PDF file of the March-April 2025 issue.

Supply chain risk just seems to be lurking around every corner right now, and it’s being matched by a host of solutions that promise to help organizations more successfully identify, address and even avoid these risks altogether. As these problem-solving applications continue to proliferate, artificial intelligence (AI) is taking center stage as yet another “go-to” tool for risk avoidance.

By continuously monitoring global news, weather patterns, geopolitical instability, and economic indicators, for example, AI helps detect potential disruptions like natural disasters, geopolitical conflicts or unexpected demand shifts. Machine learning algorithms can also quickly assess historical supplier and carrier performance data to help companies identify the most reliable suppliers and efficient transportation routes. 

The AI-supply chain risk connection, it seems, really couldn’t come soon enough. According to a recent Supply Chain Intelligence Report from logistics real estate firm Prologis, 66% of the 1,000+ executives surveyed are “losing sleep” over supply chain issues right now. And, 86% say economic instability and geopolitical concerns are influencing supply chain decisions.

“Global supply chains face an unmatched ‘polycrisis’—a web of challenges that span geopolitical tensions, economic instability, shifting regulatory pressures, fluctuating customer demands and pressing climate challenges,” Prologis points out in its report. Just 40% of executives say they have the tools, resources and strategies to tackle cybersecurity attacks, technological disruptions and regulatory changes. And, preparedness drops even further for issues like geopolitical instability, trade wars, and climate crises.

“Lack of preparedness doesn’t just heighten risk,” Prologis points out, “it directly compromises organizational performance and the ability to adapt to unforeseen disruptions.”

Suzie Petrusic, a risk expert in Gartner’s Supply Chain Practice, sees more companies making proactive moves to avoid supply chain risk right now and says the issue is top of mind for most organizations. They realize that risk and volatility have become the new normal, and being prepared in advance for catastrophic disasters like the California wildfires, economic issues like higher import/export tariffs, and geopolitical problems like the Ukraine war can pay off in both the short and long term.

For example, Petrusic says companies are using buffers and backups to ensure that their networks continue to operate when the inevitable happens. That way, the inventory is in place to hold the organization over until it can get a more permanent response in place. This is just one of many ways companies are working to avoid issues like the pandemic-driven disruptions.

“Most organizations are trying to become more proactive at this point,” says Petrusic, “because they’re feeling the cost and operational performance impacts of constantly responding reactively.”

A new ally on their side

As companies work to get out in front of the next supply chain risk they have a new ally in their corner. By analyzing real-time data and simulating various scenarios, AI helps predict potential disruptions and bottlenecks and empowers companies to make informed, data-driven decisions (e.g., where to put inventory, which suppliers are meeting service level agreements, which transportation providers are performing up to snuff, etc.).

Patricia Riedl, Americas supply chain and operations lead at Accenture, says AI can help drive down supply chain risk on two different fronts: making better predictions about what might happen and then responding faster—and autonomously—to these issues before they boil over into major problems. These two capabilities usually work in tandem. An AI-enabled digital twin replication, for instance, can map out all plant nodes, distribution centers, labor pools and product flows. It can then be used to stress-test the end-to-end supply chain.

“You can either start from the end of the supply chain and trace everything back to the consumer or vice-versa to understand what the connected impact of a disruption is across the supply chain,” says Riedl. From there, companies can use AI to build out a quick-response operating model that incorporates data-driven, autonomous responses that don’t require any initial human reaction. From there, team members can use established governance models to address issues like incomplete data sets.

Improving overall supply chain outcomes

With its knack for ingesting and making sense of large amounts of complex data, AI is also helping companies improve overall supply chain outcomes. For instance, Riedl says demand planning is one area where AI supports better decision-making and risk reduction. “This is where AI first came into play in the supply chain space,” she says. “It made sense because predicting demand is extremely difficult, and the more external data that planners can bring into forecasting algorithms, the closer it gets to ‘what you think you’re going to sell where and when.’”

Take the beverage producer that wants to target beer consumption levels right down to very specific geographies during the summer months, for example. To make the best predictions it needs to know expected weather conditions, competitive price points and expected sales volumes. The company also needs to know what major events (i.e., local festivals, concerts, etc.) are taking place in each geography.

“All of that data can be brought into a set of AI-driven algorithms that deliver a micro-targeted view of demand,” Riedl says. Then, the company can use that intelligence—and, even more AI-enabled solutions—to determine specific volumes of inventory to each specific geography. This level of precision would take months or longer for someone to handle manually using myriad different internal and external sources, but AI can handle the heavy lifting in a fast and efficient manner.

Up next, Riedl expects more companies to start using agentic AI systems to run their supply chains and address the associated risks. This type of AI exhibits goal-directed behavior and adapts its actions to achieve outcomes in dynamic environments like supply chain. “Eighteen months age we weren’t even talking about agentic AI,” she says, “but now we have several examples of companies using it to drive the inefficiency out of manual processes. We expect to see more of that.”

Look outside your four walls

Calling supply chain risk a “board-level dilemma,” interos.ai CEO Ted Krantz says visibility remains a major pain point for companies that don’t really know what their extended supply networks (e.g., Tier 2 and down) are doing. This lack of oversight can create substantial risk in a world where everything from weather events and factory fires to conflict minerals and cyberattacks can derail even the best-laid supply chain plans.

According to interos.ai, more than 480 S&P 500 firms are directly linked to high-risk regions; 3.3 million global firms have cybersecurity threat exposure; and 20 million companies will be impacted by catastrophic events this year alone. Krantz sees AI as a facilitator in the race to more effectively assess and mitigate these and other risks. However, he says these systems must factor in both internal and external data, the latter of which can be difficult to identify, gather and assimilate.

For example, Fortune 1000 companies have an average of 17,500 direct suppliers and 1.5 million relationships, according to Krantz. Knowing what all of those entities are doing at any given point is nearly impossible. AI can step in to help, but only if it’s properly primed with the right data. “Forming an opinion around decision support in a supply chain ecosystem requires market relevance—both public and proprietary,” Krantz says, “and not just your data and what you know about your supply chain.”

Understanding the limits of the technology

As more companies start using AI to ingest and sort through massive amounts of data, risk identification is becoming an important use case for the advanced technology. The related applications help organizations determine what’s most important, and then puts math and analytics into predicting geopolitical problems, supplier failures, ESG risk, tariffs or whatever else is being thrown at global supply chains at any given time. Most importantly, AI reduces the time it takes to identify, assess and react to these and other events.

“AI gives companies the power to sense what’s happening very early, and alerts them about issues like supplier insolvencies before those types of issues turn up in the news,” says Petrusic. Some organizations are also leveraging AI to help map out their supply chains. This particular kind of technology works best for organizations that have second- and third-tier supply bases, she explains, and lack a good understanding of what’s happening in those segments of their supply chains.

“AI can help them understand what that map looks like by pulling in data from different sources, and then sorting that data and identifying what’s important,” says Petrusic. Armed with that information, companies can gain a better understanding of their current state and potential supply chain risk.

And despite what you may have heard, Petrusic says what AI can’t do is predict the future. That means 100% accurate predictions around emerging risks are still a guessing game, at least for now. “If you have a relevant data set around a supplier failure, then the AI-enabled technology is going to be extremely helpful, but I don’t know of a technology that can warn you in advance about a catastrophe like the Maryland bridge collapse,” she concluded. “There are still limits around what the technology can really do for you.”

SC
MR