Beyond the Hype: How Gen AI Is Quietly Transforming Supply Chains
From smarter dispatching to AI-powered copilots, supply chains are entering a new era - if leaders act deliberately.
Not everything in supply chain needs Gen AI. But for the parts that do, the results can be dramatic.
That’s the core message from McKinsey’s deep-dive conversation on how generative AI (Gen AI) is reshaping operations - featuring experts like Knut Alicke, Alberto Oca, Asaf Somekh, and special insights from Samsara CEO Sanjit Biswas. In this episode of McKinsey Talks Operations, they unpack what’s working, what’s not, and what supply chain leaders should do next.
Why Gen AI Feels Like the Shipping Container Moment
“People overestimate disruptive tech in the short term—and underestimate it in the long run,” said Knut Alicke. He compares Gen AI’s emergence to the invention of the shipping container: not flashy, but fundamentally transformative.
Used right, Gen AI enables decision-making and execution at scale—from smarter routing to real-time warehouse optimization. But it’s only valuable when paired with foundational AI, smart architecture, and clear human oversight.
Where It’s Working: Use Cases With Real ROI
Alberto Oca highlighted how logistics players are already seeing millions in savings through targeted Gen AI deployments:
Virtual dispatcher agents that support drivers with routing, roadside assistance, and real-time coordination.
Smart co-pilot tools that connect drivers, dispatchers, and customers in one SMS thread—streamlining last-mile delivery.
Chatbots for site managers that pull data from multiple systems to answer complex operational questions on the fly.
One last-mile operator with 10,000+ vehicles saw over $30 million in savings with a $2 million investment. Another carrier saved $3.5 million with only 150 vehicles.
But not every challenge needs Gen AI. “Don’t force it,” said Asaf Somekh. “Traditional machine learning may be better in many cases.” The key is creating a flexible AI ecosystem—what McKinsey calls the AI Factory—that allows you to scale and mix models based on cost, complexity, and accuracy.
Rethinking the Role of the Planner
In one real-world example, Gen AI helped uncover inconsistencies in how order managers prioritized customers during bottlenecks. Not only did the AI flag misalignment with company strategy—it coached human planners on how to improve.
This isn’t about replacing humans. “You’re training Gen AI like a new team member,” said Alicke. The best results come when teams see Gen AI as a colleague—not a threat.
Still, getting buy-in from experienced planners used to fax machines is part of the challenge. That’s why pilot programs and curiosity-driven training matter. As Alberto put it, “It’s not about complexity—it’s about showing the value early.”
Don’t Wait to Scale - Build Smart from Day One
Many companies hit a wall after proof-of-concept. Why? They didn’t think about scale early enough. GPU costs skyrocket, applications become fragile, and integration stalls.
The advice from McKinsey is clear: treat scale as the goal from day one. Bring in your AI Factory architecture early, build with production in mind, and centralize guardrails to manage risk. Especially when using Gen AI on-premises for regulated data environments.
What's Next: A Vision of Supply Chain Agents and Copilots
The near future? Widespread deployment of Gen AI copilots embedded across planning, dispatch, customer support, and warehousing. As capabilities mature, agentic AI—autonomous teams of bots working together—will start taking on repetitive coordination tasks.
But none of it works without a strategic approach.
Start small, but think big. Identify high-value use cases in a single domain.
Design for production, not just pilots.
Mix and match AI types—don’t default to Gen AI.
Focus on workforce adoption. Curiosity, not fear, drives success.
“Gen AI won’t solve all your problems,” said Alberto. “But it will reshape how you solve them.”
What’s your view? Are you piloting Gen AI in your supply chain? What roadblocks—or breakthroughs—have you encountered so far? Share your experience in the comments.
 | A guest post by
|