We have all learned a thing or two in the past five years since the COVID-19 pandemic hit—and since that time things continue to change fast and furious in the supply chain. We have rising costs, global shocks, fragmented logistics, climate risks, and increasing customer demands. The numbers always tell an interesting story.
We see 91% of customers demand seamless end-to-end service and yet 75% of logistics leaders admit they have lagged on digital transformation and innovation. There is a disconnect here.
The reality is supply chains can no longer operate the way they did 20 years ago. If the last five years taught us anything, it painfully revealed just-in-time delivery sounds great on paper, until there is a global pandemic, or a wildfire, or a geopolitical conflict, or another unpredictable occurrence that sends the supply chain into a tailspin.
To illustrate this point, a 2024 MIT report, In Defense of Redundancy: the overconcentrated modern economy and the dangers it poses, suggests when you fine-tune your systems of production down to a set of small actions, you manufacture vulnerability. Simply, it concludes the old systems of efficiency and concentration may not fit with the new model of work.
So, where does this leave us?
In a recent conversation with David Esposito, industry advisor for logistics, transportation, and supply chain, Microsoft Manufacturing and Mobility, on The Peggy Smedley Show podcast, he says it has to be a different approach than the lean era mindset. Supplier fragility reveals there is an overconcentrated risk.
“Visibility alone isn’t enough,” he says. “Companies need risk quantification and predictive analytics to prioritize mitigation strategies.”
It is James Reason’s Swiss Cheese Effect, which is a model often shown as individual layers or slices of cheese. Absent or failed barriers at each level are represented as holes in the cheese. When holes across each level of the system line up, they provide a window of opportunity for an accident or harm. This model is used to guide root cause analyses and safety in many industries such as supply chain and manufacturing.
Ultimately, designing resiliency by default is essential in terms of implementing just in time. The question then becomes: Can AI (artificial intelligence) agents help eliminate those “holes” that exists in today’s supply chains? Esposito of Microsoft says yes.
Enter AI Agents
On a recent podcast of The Peggy Smedley Show, Esposito and I discuss how supply chains can leverage AI agents to help with contingency planning, supplier diversification, and risk management, just to name a few. He explains how we can leverage technology from freight to foresight, and if we combine with people and process, it can lead to great results. Some examples he gave include:
- Proactive contingency planning
- Scenario modeling and risk simulation James Reason’s Swiss Cheese Effect
- Predictive and prescriptive planning
- Automated safety, stock, and inventory optimization
The benefits that come along with such technologies are significant. AI-powered innovations could reduce logistics costs by 15%, optimize inventory levels by 35%, and boost service levels by 65%. In the next two decades, AI adoption in logistics could generate between $1.3 trillion and $2 trillion per year in economic value. The opportunities are simply too hard to ignore—and the risks are too high, well, to risk.
Esposito points to several case studies in the supply chain. Consider the case of C.H. Robinson, which receives tens of thousands of emails from customers daily for routine tasks. This logistics company used Microsoft Azure AI Foundry and Azure OpenAI in Foundry Models to build generative AI tools that automate emailed customer requests. The speed to market went from hours to just seconds.
In another example, materials science company Dow has up to 4,000 billing statements move through the system daily. Every line item needs verification. This is where an AI agent can enter the equation—in this case two different types of AI agents. One monitors incoming email for attached PDF invoices and structures the data for analysis, scanning for billing inaccuracies. Then, the second prompt-and-response agent can investigate further by dialoguing with the data in natural language. The result? Saving millions of dollars on shipping costs in the first year.
Of course, these are only a few examples. The bottomline is the supply chain of yesteryear simply won’t work as we look down the road in the future and certainly not 20 years from now. Supply chains can no longer hope for resiliency. They must prepare for it. The preparation starts now.
Want to tweet about this article? Use hashtags #IoT #sustainability #AI #5G #cloud #edge #futureofwork #digitaltransformation #supplychain #AIagents #logistics #manufacturing #transportation #riskmitigation