CH Robinson deploys AI agent army across ocean, air and customs

CH Robinson's AI pricing agents have already delivered more than 80,000 quotes to over 3,600 customers — and cut quote turnaround from two to three days for ocean freight to two to three minutes. That's the headline number in the US forwarder's latest push into agentic AI, which now stretches across ocean shipping, airfreight and US customs brokerage, with more than a dozen agents working in tandem on a single booking.

The numbers land alongside a telling admission. Chief technology officer Mike Neill said a decade of digital booking forms and data-exchange setups with customers achieved only about 5% automation of ocean and air booking requests.

"Why was that such a tough hill to climb?" he asked. "Primarily, because you quote your customer, then it's their suppliers booking with you. A customer might have 50 suppliers. With our agentic AI, all those booking requests no longer need to be handled manually."

CH Robinson didn't say what automation rate it now expects. Mr Neill conceded there's more to do: "We have more work to do automating all the steps for booking, but this is already a major leap forward."

What does this change for shippers and forwarders?

For shippers, the pitch is speed when it matters — securing capacity or re-planning around disruption. "When plans change or disruptions occur, customers can more quickly evaluate a new routing, mode of shipping, schedule, or price," said Mike Short, president of Global Forwarding.

The agents do more than pull a rate. They interpret shipping requirements, chase information from external parties, and assess routing options and charges at both ends. Mr Short used Shanghai as an example: the choice of port and terminal changes inland transport costs and fees, and terminal fees move constantly. One agent now checks terminal websites daily and monitors emails for fee notices, so — in his words — people in two different time zones don't have to call every terminal for every quote.

Where a customer hasn't specified a port, agents can propose gateways, inland transport by truck, rail or barge, and consolidation centres.

"Customers have information they need sooner. Data is more consistent. Processes are more predictable," said Mr Short, arguing that "the benefits keep compounding" as the AI playbook moves into global forwarding.

Rollout timeline and customs gains

• Ocean booking agents launched in China in August
• Rest of Asia follows in Q4
• Global rollout expected in Q1 2027
• Pricing agents, introduced for North American LCL in 2025, now cover FCL and airfreight across several regions

In US customs brokerage, AI and automation classify documents, extract data and request missing paperwork. CH Robinson claims a 23% reduction in time spent on the affected work, with human brokers focusing on validation and compliance.

The moat argument

Mr Short made the boldest claim of the announcement: "With our maturity in using AI to augment our talented experts, only CH Robinson is equipped to transform something as complicated as freight forwarding."

Chief strategy and innovation officer Arun Rajan was slightly more precise about why the company thinks rivals can't copy it. "Our AI agents continue to operate on 100 trillion data points and a context layer of customer-specific information that simply cannot be replicated elsewhere," he said. The company's 450 in-house software engineers and data scientists, who built the tooling for trucking and 4PL, are now pointed at widening the competitive moat in global forwarding.

The financial evidence is suggestive rather than conclusive. Q2 forwarding operating profit rose 18.8% year on year while average headcount fell 16.6% — but those figures cover an earlier stage of automation and don't isolate AI's contribution. To put the headcount move plainly: the division is doing meaningfully more with about a sixth fewer people.

Rivals will note what wasn't said. No target automation percentage, no revenue attribution, no timeline for the customs technology beyond its US deployment. For now, the claim rests on quote latency and agent counts — real metrics, but ones any large forwarder with a data lake will now be racing to match. The Q1 2027 global ocean rollout is the date that will tell whether the moat is real or just well-marketed.