There are two numbers that might help define the impact of AI at 3PL giant C.H. Robinson.
One of them is 55.3%. That’s the growth in the company’s stock price during 2025. There are no other logistics companies that had a record like that in the recently-completed year. Every analyst would say AI is a key part of that surge.
The second is 30. That’s the number of actual agentic AI tools in use at C.H. Robinson.
Damon Lee, the company’s CFO, reviewed C.H. Robinson’s use of AI in a recent interview with FreightWaves. Lee’s comments moved away from the general message on AI usage that C.H. Robinson (NASDAQ: CHRW) has been touting–with profitability and operational figures to back it up–to more specific details on what the company is using AI to get done.
It’s impossible to know if 30 is a large number or a small one. There is no measuring stick for how many agentic AI tools a freight brokerage that will produce 2025 revenue somewhere close to $11 billion should have at this stage in the AI revolution.
But the verdict from investors seems to be that 30 is a good number, they like the results and are hoping for more in 2026.
Lots of companies read invoices
At the annual meeting of the Transportation Intermediaries Association in April, a parade of companies touted their freight tech applications at a day-long session with journalists. Most of them talked about AI, but the message at one point became sort of repetitive: we use AI to read invoices or take an incoming call and convert it to language-based data that a broker could use.
The reality is that many companies across the spectrum of industries–not just logistics–that have plunged into AI aren’t seeing benefits, at least not yet.
That isn’t surprising to Lee. “I believe people using an off-the-shelf solution for AI, in many cases, will be nothing but a cost adder for their business,” he said. “I don’t think they’ll ever see positive productivity to offset the cost that they’re incurring, because the pay-by-the drink model of AI is very expensive.”
450 men and women at work
That’s why C.H. Robinson has a team of 450 engineers writing their own AI applications. They are the source of the 30 agentic AI tools cited by Lee.
“We’re using bespoke customized AI solutions to drive demonstrable business results,” Lee said.
Lee spoke about one of the 30 agentic AI solutions in going from the theoretical to the specific.
C.H. Robinson receives requests for about 600,000 rate quotes each year at its North American Surface Transport (NAST), the division that includes its core over the road brokerage activities.
Historically, it had the ability to respond to about 60% to 65% of those quotes, Lee said.
“So when a person was manually doing that exercise, somewhere around a third of those requests either never got answers, or they got answered in a timeline that didn’t meet the customer expectations,” Lee said.
With an agentic AI tool now at the center of the process for responding to those requests, Lee said C.H. Robinson is able to respond to 100% of the queries.
“I’m now getting to a third of the universe of freight that was available to me before that I never got to,” Lee said. “The sophistication in how I’m responding to the customer has gone up exponentially.”
A human might have five to 10 data points to use in responding to a quote. But the agentic AI tool, Lee said, has “tens of thousand if not hundreds of thousands of data points available to them.”
The price quote from the human would be “unsophisticated,” Lee said. Response time was 17 to 20 minutes.
But Lee said the agentic AI tool will respond in 32 seconds.
Optimizing margin rapidly
Asked to describe “favorite AI agent number two,” Lee turned to a tool he said is designed to optimize revenue management.
Pricing strategy at C.H. Robinson, Lee said, would be “fairly unsophisticated. It was targeted to figure out ‘I want to try to get this much margin, or this much volume, at this much margin.’”
If the gross margin rate in that strategy is 40%, loads are accepted or rejected on that basis, Lee said. “At the end of the month or the quarter, I see how my pricing strategy did,” he said.
What he described as “course correction” in the middle of that strategy wasn’t easy, according to Lee.
With the strategy now being powered by an in-house developed agentic AI tool, “we’ll set a pricing strategy at 8 a.m. on Monday, and by 8:05, we’re testing that strategy,” Lee said. The tool asks the question, “is the strategy yielding the volume and the margin that I expected to get?”
And if the answer is no, Lee said, “it will change the strategy two minutes later.” That replaces a system, Lee said, where “you might have a 30-day increment with no adjustments, or even a 90-day increment with no adjustments.” In its place, Lee said, is an approach where “the strategy gets adjusted hundreds of times a day.”
Lee referred to that model as “gross margin arbitrage.” Previously, Lee said, playing that strategy was largely impossible.
“Folks would say gross margin is commoditized,” Lee said, defining gross margin as a formula of revenue per load minus the cost of capacity. “But we’re able to optimize price, and we’re able to optimize cost because of the frequency in which we’re able to test the market and react to the market in terms of price,” he said.
If the data coming from the agentic AI tool is showing “volume coming in hot, I can choose to optimize margin,” Lee said. But if “there’s not as many loads on the market, I can get more aggressive in price.”
That’s what every broker does. But C.H. Robinson’s message is they can do it far more rapidly because of AI.
C.H. Robinson does not disclose gross margin in its earnings. It’s a derivative of adjusted gross profits, which are published with the company’s earnings.
The most recent earnings report shows that in the three months ended September 30, C.H. Robinson’s adjusted gross profit in its truckload brokerage was down 2% year-on-year. That is a relatively small drop considering the freight market in 2025 compared to the already weak one in 2024.
But the 3PL’s adjusted gross profits in its LTL operations were up 10.5%. For the nine months, the figures were down 0.9% in truckload and up 6.7% in LTL
Not everybody is buying the C.H. Robinson story. For example, data on Yahoo Finance shows a 6.47% share of the company’s stock float was sold short as of December 15. That is a relatively high number.
The question is whether C.H. Robinson’s price has surged because it’s a brokerage play–hard to imagine that, given the performance of other 3PLs like RXO (NYSE: RXO) and Landstar (NASDAQ: LSTR)–or an AI play. Lee wouldn’t say it wasn’t because of brokerage, but he said “we’ve had a few investors tell us we may be the only company that is getting the level of benefits from AI that we’ve enjoyed in this ecosystem.”
Within the AI play, Lee said there are chip makers and data centers and other companies whose stocks are more of a pure AI play.
What are harder to find are the operating beneficiaries, Lee said, “examples of folks that are winning at the application layer. And that’s certainly where C.H. Robinson is winning.”
More articles by John Kingston
New York City bill that targeted Amazon won’t get taken up in 2025
State of Freight: a depressed trucking market suddenly comes to life
Benesch panelists: Why 2026 could be a strong year for logistics M&A
