The march of artificial AI into the world of logistics is accelerating, specifically around the development of AI agents built directly into standalone software or tools provided by third-party logistics providers (3PLs).

Those advancements are impacting key processes in both the international and domestic logistics environment, including demand planning and auditing freight bills, but more broadly agents are popping up in virtually every software platform with which shippers interact.

Yet, there remains confusion and residual skepticism about the extent to which AI agents can transform logistics workforces. Most logistics professionals have a sense they are either already using AI agents or interacting with them in an unseen way but aren’t yet experiencing a different way of working.

As more major logistics players make investments in agentic AI, that could soon change.

The latest example is C.H. Robinson, which on Monday said it has enhanced its ability to use AI to uncover data from sources where it has been traditionally hard to extract and categorize information.

C.H. Robinson, the largest US freight broker and a top 10 ocean forwarder, has leaned heavily into AI over the last few years as a means to reshape its business.

“With agentic AI, we’re unlocking the value trapped in unstructured data: phone calls, emails, tribal knowledge,” the company’s chief technology officer, Mike Neill, said in a statement Monday. “In September alone, one of our AI agents captured 318,000 freight tracking updates from a single type of phone call. Previously invisible to our systems, that data now flows to another AI agent that updates our platform, feeding our predictive [estimated times of arrival] and optimizing our customers’ deliveries.”

While C.H. Robinson is a prime example of the largest companies in logistics recrafting their service offerings around agentic AI, it is far from alone. Among other large providers, project44 and FourKites have built entire sets of AI agents to support their ambitions to appeal more broadly to shippers.

WiseTech Global — provider of the most widely-used freight management software among global forwarders, told the Journal of Commerce in September it is similarly building an array of AI agents to allows its customers to better utilize its customs, tracking and accounting products.

And Maersk launched an AI customs tool in June designed to benefit from the large datasets that the shipping line’s scale provides. C.H. Robinson, with the largest dataset in truckload brokerage, has made a similar connection.

Wide spectrum of tools

The activity is far from confined to larger players in the market. Last week, ShipAngel, which started as a freight rate management software vendor for shippers, announced it is now offering a more comprehensive set of tools for shippers, all driven by embedded AI agents that can book and track shipments off an already established rate.

“Enterprise software is undergoing a once-in-a-generation transformation,” ShipAngel CEO Graham Parker said about what the company is describing as a shift into enterprise resource planning (ERP) specifically built for supply chains. “Monolithic systems are giving way to modular, intelligent architectures.”

Parker’s contention is that most shippers are burdened with ERPs that are built mostly for finance and planning teams, not supply chain teams. ShipAngel’s use of AI agents — also known as digital workers — is key to its ability to gather unstructured data that’s often difficult for those teams to gather and make decisions about.

In late September, Jason Traff, CEO of transportation management system (TMS) provider Shipwell, told the Journal of Commerce Inland Distribution Conference that the most likely functions to be handed to AI agents are demand planning, truckload scheduling, freight bill auditing, and logistics cost analysis.

Traff stressed that AI agents’ strengths lie in adapting to new variables and simulating human-like behavior and reasoning. AI-based workflow automation, on the other hand, is strongest at pattern recognition and handling sets of complex rules.

Accessing unstructured data

At the Virginia Maritime Association’s (VMA’s) International Trade Symposium in Norfolk Thursday, Vijay Harrell, CEO of software provider TradeLanes, described how AI agents within the company’s product allow its exporter customers to get automated messages around vessel schedule amendments or changes to earlier return dates. Those changes have historically been detected by having people — at the exporter or the exporter’s 3PL — periodically check carrier or terminal websites.

Paul Albert-Lebrun, CEO of container terminal software provider Nodal, similarly explained on an AI panel how digital workers would help terminal operators get more out of their existing terminal operating systems and extract key pieces of unstructured data from adjacent systems to make informed decisions on things such as labor deployment and crane operations.

At the RateLinx Insight Conference Tuesday in Scottsdale, Arizona, CEO Shannon Vaillancourt described how AI helps shippers make informed decisions that have traditionally been negatively affected by human sentiment. For instance, he said people tend to “remember things that didn’t happen and anchor decisions based on irrelevant data.” He also said people resist change because they fixate on bad experiences with transportation vendors that aren’t representative of that vendor’s objective performance.

But Mac McCall, a senior research associate at the Virginia Tech Transportation Institute, told the VMA conference that a future based around digital workers need not exclude people.

“I’m a Star Trek fan, and if you look at the bridge of the Enterprise, you see the most advanced technology available, but there are still people being presented with information and making decisions,” he said.