AI Is Eating Logistics

Petersen says Flexport aims to make ocean container shipping 8–10% cheaper over the next few years. Its “scale economies shared” model passes some savings from automation and greater volume to customers, which may encourage them to ship more with Flexport.

Petersen says established companies have useful data, domain knowledge, and customer relationships, while Flexport also controls its own technology stack. He describes hackathons where about 90% of recent projects used large language models, and a program giving non-engineers one day a week for 90 days to learn AI tools and automate parts of their work.

A hackathon project lets customers ask questions in natural language and get reports without writing SQL or building dashboards. Flexport’s planning system saved 2% of its ocean freight spend while improving transit time by 20%, in part by finding replacement space when bookings are canceled and moving containers to earlier sailings.

Petersen sees language models working alongside existing optimization tools, for example by asking customers whether a shipment can move earlier. Flexport also uses agents to turn booking emails into orders, confirm warehouse details by email or phone, and flag potentially unhappy customers. He says the company expects automation to cover 90–95% of its work as AI improves.

Petersen wants customers to arrange shipments anywhere through code or voice while Flexport manages the work between factories and consumers. The company shipped to and from 147 countries but had employees in 22; its stated plans were to cover 95% of container trade by 2028 and operate wherever legally possible by 2035.