Pax uses AI to scale drawback
Penny Chen, CEO of Pax, on building AI-powered tariff refunds
The key change is not better refund math, it is a lower minimum claim size that turns drawback from a niche enterprise service into a broader software market. Traditional providers spend months cleaning invoices, bills of lading, HTS codes, and export records by hand, so they need large refunds to justify the labor. Pax is using AI for document extraction and software for claim assembly, which makes smaller accounts economic to serve.
-
The bottleneck is paperwork, not eligibility. Drawback can return up to 99% of duties on qualifying exports, but every claim has to be filed electronically through CBP systems, with detailed import and export records matched correctly. That creates a fixed workload even before refund size matters.
-
Legacy firms are built for large accounts. In this market, Charter Brokerage processes about $1B in refunds and holds roughly 30% share, which shows how concentrated drawback has been among specialist providers serving high volume shippers with enough refunds to cover manual work.
-
The closest comparable is Flexport, which is also using AI inside customs workflows and says its duty drawback product delivers 20% to 40% higher returns than traditional tools. That suggests the wedge is becoming software plus operations, not pure brokerage labor.
This points toward drawback becoming embedded into everyday trade operations for mid market importers, exporters, brokers, and 3PLs. As filing, data cleanup, and refund optimization become more automated, the winning companies will look less like consulting shops and more like trade infrastructure, with drawback as the first product in a larger customs and compliance software stack.