C.H. Robinson Expands AI Agents Across Shipment Life Cycle

C.H. Robinson has expanded its proprietary fleet of artificial intelligence agents to handle more tasks across the life cycle of a shipment, the company announced April 16.

The company launched the generative AI agents in 2023 to help automate the shipping process. This followed more than a decade of C.H. Robinson working with AI technologies and bringing what it learned to the service. In that time, the company has been able to leverage shipping data to create new agents informed by more complex and dynamic models.

“You have to build your lower-level agents first to do your basic tasks,” said Mark Albrecht, vice president of artificial intelligence at C.H. Robinson. “Then these new reasoning agents, they’re best placed a level above, where you look at what employees do, a complete circle of their tactical work in a day, and then how to orchestrate the lower-level agents that you’ve been building that do things like build out orders, set an appointment. How could you orchestrate across those to create a cohesive workflow at the employee level to help them?”

The AI agents can perform numerous tasks like price quoting, processing orders, acquiring trucking capacity, setting delivery appointments and supply chain tracking. The earlier models were first applied to emails to extract information about available trucks before expanding their capabilities. The company is also extending these tools to the small and medium-size business sector. These agents handle basic tasks, freeing workers to focus on more valuable work such as helping customers.

“All that information was in people’s heads and on emails, and we use the AI to get that into a structured format,” said Megan Orth, senior director of connectivity at C.H. Robinson. “And so now those agents are able to get this data into structured formats that then allow our next generation of agents to be even smarter with that data.”

RELATED: Generative AI Driving Freight Brokerage Transformation

Orth added that these newer agents have low-level reasoning capabilities, while the older models were more direction-oriented. A key difference is that the new agents can check their own work to produce better responses. They might also draw attention to a data point that could be relevant even if not strictly asked to look for that information.

“With these newer agents, we’re able to set up a framework much quicker than we ever could and tackle that problem,” Albrecht said. “Because you can just say, hey, your job as an agent is to determine a freight class. It knows what that means because it understands the internet and knows what freight classes are. You also give it a tool to search a database of these hard commodity classes. With the older agents, you couldn’t do that. Now it has tools so it can look and do a search in the world and find things that a person would through a search process. And then, because it’s a reasoning agent, it can check its work quickly and come to a determination.”

C.H. Robinson noted that its AI agents have already performed more than 3 million tasks. The new agents that now sit on top of the technology structure improve on earlier, lower-parameter models by leveraging contextual reasoning capabilities to allow for more complex and dynamic tasks.

“We’re at well over 1 million price quotes delivered by AI,” said Arun Rajan, chief strategy and innovation officer. “In March, we hit 1 million orders processed by AI. Those are two of our most mature generative AI agents, and they’re more capable every day as the models we’ve built get smarter and as we apply them to more of our 83,000 customers. Each additional shipping step we’ve automated beyond those has created new leaps in efficiency for global supply chains and freed our people to do more high-value work for our customers.”

C.H. Robinson ranks No. 2 on the Transport Topics Top 100 list of the largest logistics companies in North America.