Analyst Insight: Generative artificial intelligence has been a compelling new development in recent years, because it introduced new possibilities for automation in an industry that requires so many process steps and information transfers just to move a single shipment. Suddenly, AI could handle information that wasn’t perfectly structured in a transportation management system, an API or web form. Because of generative AI’s ability to mimic human cognition, it meant that business transactions handled through email, phone calls and ordinary documents could be automated, too.

The emergence of agentic AI has been an even bigger leap forward for logistics. These systems don’t just read and analyze information or generate text. AI agents can make decisions in the process of performing tasks, acting autonomously to achieve defined objectives.

Already, AI agents are performing shipping tasks such as delivering price quotes, processing orders, setting appointments, getting tracking updates, and performing a multitude of other essential steps in the lifecycle of a shipment. They’re executing increasingly complex actions, and they’re superb multi-taskers, capable of processing 20 orders simultaneously in the same 90 seconds it would take to do one. AI agents are even assisting other AI agents.     

This has allowed a new premium level of logistics service. Since AI agents are always on, shipper needs can be attended to just as well at 1 a.m. as at 1 p.m. Because AI agents act in seconds instead of hours, they’re securing more favorable rates, carriers and appointments for shippers. And by repetitively performing millions of tasks, they can deliver great precision and reliability.  

One challenge for many organizations is building logistics-specific AI models, so their agents are fit for purpose. Just like people, AI agents need expertise in their specific domain in order to accurately reason and make intelligent decisions. No one wants to trust their supply chain to a novice.

Another obstacle for those still experimenting with AI is having the scale of data to properly train AI agents. Every shipment is different, so any agent will only be as smart as the breadth and depth of freight and shipping conditions reflected in the dataset behind it. The bigger the dataset, the more “brainpower” for the agent. Some companies will find that too small of a dataset means AI agents simply can’t deliver the hoped-for results or return on investment.

Resource Link: https://www.chrobinson.com

Outlook: Looking ahead, expect AI to have an even broader impact in logistics, from planning to replenishment. AI will also likely act more predictively and proactively, and predictive insights and recommendations will enable more control over supply chains, and allow for smarter cost optimization. Always-on service will allow instant response to shifts in demand and shifts in market conditions. Systems will anticipate disruptions and proactively reroute freight. 

Today, we’ve entered the era of the agentic supply chain, where shippers can have a self-optimizing supply chain that continuously thinks, learns and adapts.