MHI Solutions Improving Supply Chain Performance
The next generation of artificial intelligence—called agentic AI—is being hailed as the technology that will finally fulfill the promise of autonomous supply chains that are smarter, faster, more agile, more efficient and more reliable than ever before. What are the real opportunities and roadblocks, and when can we expect agentic AI to make a real impact? We talk to experts who separate the wheat from the chaff.
Oliver King‑Smith, co‑founder and chief technology officer for MHI member sSy.AI, recalls a few tense moments during a live demonstration of the company’s Supply Chain Operating Technology Intelligence platform (SCOTi) when the CTO for a prospective buyer asked it to identify bottlenecks in his operation.
Pronounced “Scotty”—a nod to sSy.AI’s Scottish links—the software promised to transform warehouse operations using a new kind of artificial intelligence (AI) called agentic AI to analyze vast amounts of data and provide insights to optimize layout, inventory management and workflow efficiency. And do it fast.
After “a bit of a heart‑stopping 30 seconds while SCOTi sat there thinking about it,” an answer came back that provided the CTO with an actionable insight: SCOTi found there was a long delay between when the pick was happening in the warehouse and when it was actually shipping out of the warehouse.
“Now it didn’t sit down and explain exactly why that problem was happening, but it explained that there was a real problem,” King‑Smith said. “There was a real delay there and it was happening in the warehouse operation.”
The story highlights just one capability of agentic AI, software programs that empower autonomous AI agents to make decisions and take actions to achieve specific goals—by themselves or in collaboration with digital and physical tools such as application programming interfaces (APIs) or robotics systems.
With those capabilities and the ability to operate in complex environments and adapt to changing conditions, experts say agentic AI will supercharge supply chains, take automation to the next level, increase productivity and resilience, cut costs and enhance customer experience and worker satisfaction.
By 2028, 33% of enterprise software platforms will include agentic AI, up from just 1% in 2024, and 15% of day‑to‑day work decisions will be made autonomously, up from zero percent today, according to supply chain technology consultant Gartner’s Top Technology Trends for 2025 report.
Top use applications for agentic AI in supply chain and logistics include warehouse optimization and inventory management, predictive maintenance for fleets and equipment, enhanced demand forecasting and process automation—all based on its strength in gleaning insights and trends from oceans of data.
Solutions providers also are working on applications of agentic AI technology that will automate functions like renegotiating contracts and other aspects of supplier relationship management and managing compliance and risk management systems with minimal human supervision.
Andrei Danescu, CEO and co‑founder of MHI member Dexory, a developer of warehouse intelligence platforms, sees agentic AI as the next step in the evolution from a passive “blind” warehouse operating on scant information to an intelligent warehouse able to adapt based on data collected in real‑time.
“This technology, in terms of AI and autonomous systems, is really becoming front and center, the driver for operational excellence in logistics and supply chain,” Danescu said. “It’s probably going to be one of the biggest revolutions in logistics that we’ve seen since the invention of the forklift and the pallet.”
Closing the ‘AI Agency Gap’
Agentic AI is looking to build off the recent advances in artificial intelligence known as generative AI or GenAI. Since OpenAI introduced ChatGPT in 2022, others have followed, including Anthropic’s Claude, Google’s Gemini, Microsoft’s Copilot and most recently, Chinese upstart DeepSeek.
These products are trained using so‑called large language models (LLMs) to perform structured tasks, generate content such as text and video and respond to relatively simple queries. Think business tasks like email response and customer support, or personal tasks like ordering flowers for your sweetheart.
While a huge step up from deterministic chatbots limited to dishing out canned responses, LLM‑based assistants like ChatGPT still leave what Gartner calls an “AI agency gap” between AI and human abilities to adapt, plan, achieve complex goals, act autonomously and continuously improve through learning.
Agentic AI promises to close that gap with the use of specialized language models trained on domain‑specific data that enables AI agents to perform tasks reliably and repeatably in a defined setting such as a warehouse, act on insights from real‑time data and interact with each other to solve problems.
For example, at ProMat 2025 in Chicago in March, sSy.AI launched a new version of SCOTi specifically for the material handling industry that allows users to customize the software down to the level of a single warehouse or DC and ask questions about their data and get actionable insights in natural language.
Eager to Leverage Technology’s Benefits
Those walking the show floor at ProMat wouldn’t have seen many technologies already incorporating agentic AI—the concept is still very much in the experimentation stage—but many exhibitors showcasing solutions there are working to build it into the next generation of their products.
A lot of the excitement is focused on how agentic AI can be used in combination with other developing technologies such as computer vision, image processing and robotics to bring a new level of autonomy and orchestration to manufacturing, distribution and logistics operations across the supply chain.
Dexory’s Danescu, for example, looks into the future and sees how agentic AI could enhance the company’s digital twin platform, DexoryView, which is updated continuously with data collected by its proprietary 14‑meter‑tall autonomous mobile robots to provide real‑time warehouse visibility.
Currently, the system relies on traditional AI and machine learning to bridge the gap between data and insights, but Danescu foresees a day when agentic AI models will be able to “spit out an agent” capable of re‑slotting inventory or regenerating warehouse layouts based on the data provided by the AMRs.
“That’s what gets us really excited,” Danescu said. “This is going to be completely transformational for the way people are actually running their businesses, for the way the supply chain actually operates at the global level.”
Raffaello D’Andrea, CEO and co‑founder of MHI member Verity, sees agentic AI’s potential to improve operational efficiencies by enabling warehouses to autonomously process and act on data from Verity’s AI‑powered drone system, which uses computer vision to find discrepancies such as misplaced inventory.
Verity’s system is deployed in more than 150 sites worldwide, including by its largest customer, IKEA. The company, in a pilot project with footwear brand On at a Maersk warehouse in California, is now testing an RFID‑based system that provides visibility beyond line‑of‑sight and has proven 99.9% accurate.
Instead of prescriptive supply chains where, for example, an employee alerted by the WMS instructs a forklift driver to move the misplaced goods, D’Andrea envisions a future when agents powered by agentic AI collaborate to fix the error on their own, saving time and freeing up workers for other tasks.
“You want to have that higher‑level agent say go and store it in the best place possible,” D’Andrea said. “And then a system like ours would determine the optimal storage location, where to retrieve it when needed, and autonomously handle the entire process—without requiring predefined instructions.”
One company hoping to leverage agentic AI is MHI member FORTNA, which announced at ProMat a new partnership with sSy.AI that will use SCOTi to develop next‑generation warehouse analytics and warehouse design to improve material handling operations in e‑commerce and omnichannel facilities.
The R&D project will include how sSy.AI’s 3‑D computer vision technology, winner of the ProMat 2023 Startup Innovation Award, can be combined with the analytics capabilities of agentic AI to convert video feeds from warehouse cameras into real‑time data that can be used to update digital twin platforms.
Analytics for inventory management, demand forecasting, route optimization and other purposes has been the entry point for agentic AI into the material handling and supply chain space, but experts say the technology’s full impact won’t be felt until what AI chip maker Nvidia calls “physical AI” becomes reality.
That will be the day robots and other systems fulfill the promise of true supply chain automation as they harness agentic AI to convert data into physical action, becoming intelligent, reasoning machines able to perceive their surroundings, make decisions on the fly and turn on a dime as conditions dictate.
“Now is the time for robots,” Nvidia CEO Jensen Huang confidently proclaimed to some 25,000 people packed into a hockey arena in San Jose on March 18 for the company’s annual AI conference. Agentic AI is the reason why.
Click here to read the full feature.