Global trade volumes keep climbing, but customs teams aren't growing with them. Qualified declarants are hard to find, and hiring your way out of the gap is both slow and expensive.
As a result, customs brokers and freight forwarders all face the same question: how do you process more declarations without taking on more compliance risk? AI is part of the answer, but only if you apply it the right way.
A typical declaration starts with a stack of documents, such as invoices, packing lists, and bills of lading. First, the declarant hunts for the right data. Then they type it into the customs system, sometimes running it through an Excel pivot table along the way.
This way of working creates four problems:
What makes it harder is that information for one declaration rarely sits in one place. Prices are on the invoice, while weights are on the packing list. Other details, in turn, are buried in transport documents or certificates.
Reading a single document is easy for AI. The real work, however, is combining data from all of them at line level and turning it into correct tariff lines. That is exactly where most of a declarant's time goes, and therefore where automation pays off most.
Many people assume AI means fully automatic. In customs, though, that's a liability.
Customs is a compliance business. Even a small error can lead to a fine or a delayed shipment, and your customer pays you precisely to avoid both. In other words, your reputation rides on every declaration.
For that reason, the right tool doesn't act as a black box. Instead, it shows where it found each value and how it got from the document to the structured data. The declarant then checks, adjusts, and approves. After all, speed only counts if the declaration is right, and if you can show why, it's right.
Goods classification is a good example of why control matters. Every good you import or export needs a commodity code: eight digits for export and ten for import. If that code is wrong, the rest of the declaration doesn't matter.
That's why pulling a code off an invoice isn't enough. A reliable system also checks against the full nomenclature, so that the code is both relevant to the goods and valid. Even then, the declarant still makes the final call.
Most people already use ChatGPT, Gemini, or Copilot at work. Customs teams, on the other hand, remain wary of AI, and the fear behind that is simple: will it replace me?
The short answer is no. What AI replaces is copying and retyping. As a result, declarants get more time for what they're trained to do: handling complex cases, advising customers on legislation and keeping operations compliant. Their expertise therefore becomes more valuable, not less.
The technology is only half the story; getting a team to trust it is the other half. Fortunately, a few principles make all the difference.
The time savings show up first. In the long run, however, the bigger win is consistency.
Rising volumes won't wait for the labour market to catch up. The good news is that AI can close the gap, as long as your declarants stay in control, and every declaration stays traceable.