News / Flexport doubles down on AI and enterprise shippers as it moves beyond start-up

Flexport founder and CEO Ryan Petersen says the company is entering a new phase focused on artificial intelligence and large enterprise customers, as the digital freight forwarder moves beyond its start-up phase. 

At TPM in Long Beach, Mr Petersen said Flexport was accelerating efforts to automate large parts of freight forwarding operations, particularly the coordination work that sits between shippers, carriers, and service providers. 

At the same time, he pushed back against the long-standing perception of Flexport as a purely digital player. 

“We are very much dealing with physical offices in more than 40 locations, covering about 85% of global trade with our own people solving problems on the ground,” he said. 

The company, he argued, was built around combining technology with a global operations network rather than replacing traditional forwarding activities entirely. Artificial intelligence is now central to that strategy. 

Mr Petersen said he had declared a “code red” internally on AI development, pushing teams to rapidly build agents capable of performing tasks that traditionally required manual work. 

One of the first areas to see major change has been customs compliance. Flexport introduced an AI auditing system last year that reviews customs entries before they are submitted. Previously, compliance teams manually rechecked only a small percentage of filings. 

“Our error rate went from 1.8% to 0.2%,” Mr Petersen said. 

The results convinced him that AI could significantly reshape operational processes across logistics. Freight forwarding still involves large volumes of “white-collar transactional labour”, coordinating shipments, exchanging documentation, and managing exceptions between shippers, carriers, and service providers. 

Flexport estimates roughly 10% of the cost of shipping a container relates to that coordination layer. 

“If you can drop the price of shipping everything in the world by about 8%, it would be huge,” he said. 

Alongside the technology push, Flexport has also been repositioning its commercial strategy. 

Mr Petersen said the company had spent the past two years moving “upmarket”, targeting larger enterprise customers rather than primarily small and medium-sized shippers. For these companies, compliance and supply chain optimisation tools are often the entry point. 

Tariffs in particular have become a major issue for many importers. 

“There were 53 changes to the tariff code last year,” Mr Petersen said, adding that keeping track of those changes had become a significant challenge for companies operating global supply chains. 

Flexport has launched several tools designed to help companies track tariff changes and identify potential refunds on duties paid.  The company recently introduced a refund calculator that analyses US customs data to determine whether shippers are owed money following tariff adjustments. 

Flexport says it has already identified around $4bn in potential refunds through the tool. 

“That’s a problem everybody has,” he said. 

The company’s evolution has also been accompanied by a shift in Mr Petersen’s management style since taking back the CEO role – he said he now takes a far more direct approach to running the business. 

“I talk to probably 150 people a week internally,” he said, adding that he had also met around 600 customers in person over the past 12 months. 

Previously, he admitted, he had relied more heavily on senior executives to manage day-to-day operations. But he said: “You shouldn’t trust everything your team tells you. Customers don’t usually lie.” 

Looking ahead, Mr Petersen confirmed that Flexport still intended to go public, although timing remains uncertain. 

“We want to go public,” he said. “But it’s a question of timing and market circumstances.” 

As automation spreads through the logistics industry, Mr Petersen believes the biggest challenge may not be competitors, but how quickly companies can adapt. 

For Flexport, he said, the priority was clear: “Can we build the AI?”