Human Element Remains Critical as Companies' Use AI for Trade Compliance, Experts Say

Customs brokers will remain relevant alongside AI because "we'll leverage the skills that propelled us to where we are as humans now," said panelist Vince Iacopella, president of government and trade relations for Alba Wheels Up.

As an example of a situation where humans are able to understand context in a way that AI cannot, Iacopella talked about how, even though autonomous driving Waymo cars would abide by the neighborhood speed limit, the humans living in the neighborhood know to drive slower down the street when children are outside playing.

"Humans will have to provide context," and that includes being able to gauge people's expressions and respond to their needs or concerns, according to Iacopella. Humans also can respond to ethical questions that may arise with trade compliance, he said.

While the human factor will remain integral, implementing AI technology into a company's trade compliance regimen does have its benefits, particularly in a policy environment that appears unpredictable, according to Sam Bosu, CEO and co-founder of Amari AI.

These ever-shifting trade policy changes open up a "huge compliance risk. There are post-summary corrections if you're underpaying duties ... and if you're overpaying duties, your customers will get angry," Bosu said. He continued, "So, with AI, the goal here is that it won't solve everything, but [it will] ease out with some of the transitions. [AI integration] not only boosts operational productivity, but you will also see [fewer] rejects and PSCs, and it should help you stay more compliant with the liquidations."

Bosu described a three-pronged framework for integrating AI responsibly: that integration needs to enable auditable results, and it needs to build trust and accountability.

For auditability, "let's say the AI came up with HTS [codes]. Can it show me what are the CROSS rulings behind it? Were there similar SKUs with similar outcomes in my previous shipment, and did it pick specific parts from either the AMS query or previous shipments or some of the commercial invoices," Bosu said. "Oftentimes we have heard the AMS query says it's 30-piece count, but my document says 20-piece count, so can AI pick that up and say there's a discrepancy for the human to go ahead and review?"

AI can also be used to understand the reasons why something may have been rejected from claiming USMCA, Bosu continued.

"As Vince mentioned, there is some period where there will be a transition from entry writers to entry reviewers" as they get comfortable with AI, Bosu said.

But "we also don't want them to be too comfortable where they have their eyes closed and are just accepting everything that the AI says, which unfortunately happens in many cases. So the last piece of it is accountability," which can be developed by comparing AI's suggestions on how to respond to determinations by CBP and partner governing agencies with those by compliance professionals, according to Bosu.

"We see brokers spending a lot more time on consultative work instead of doing paperwork," Bosu said.

Indeed, as companies integrate AI into trade compliance operations, customs brokerage is experiencing a generational change, according to trade consultant Tom Gould. That's because the government also is using AI to enforce regulations, including verifying information and looking for unusual patterns, and so companies must also keep pace, he said.

"So much of what's happening is changing what we're doing. And to think about what we're doing, we need to think about what we need to do to be relevant in the future. And what we need to do to be relevant in the future is shift and move away from being customs brokers ... to [being] advisers," Gould said. "I see us as customs brokers becoming much more professional and much less clerical. And if you're able to embrace that, if you're able to become that adviser to your customers, I think you're going to be relevant and remain relevant and become extremely relevant in the very near future."

Iacopella said integrating AI into a company's trade compliance regimen can also build upon the broker's duty to exercise responsible supervision and control when exercising oversight over customers, transactions and compliance with trade laws.

"We see regulatory compliance, and we have it under regulatory compliance, but really, it's really back-office efficiency," Iacopella said. AI tools can help brokers discover missing data as the tools analyze high volumes of PDFs and XML files or EDI files, which in turn can ultimately enable better regulatory compliance and risk assessments, he said. Brokers can also take all this information and use it in a way to advise their clients on how they can be compliant.

As multinational companies grapple with responding to trade compliance regulations of multiple countries, AI can be used to build similar workflows within a multinational organization, Gould said.

"I think it's important to think about the fact that AI doesn't have nationality," Gould said, noting that U.S.-based customs brokers are focused within their jurisdictions. "We're going to have to think much more globally as we adopt AI, because if we have a workflow process in here, in the U.S., your counterparts in another country are going to have a very similar workflow process. And why build multiple AI technology tools to manage those workflow processes differently? Why don't we start thinking internationally?"

Said Basu: "I agree with what Tom has said about the majority of the workflows being pretty similar. I would expect 90% of the workflows internationally to be the same. And I would encourage, rather than having a year or two-year timeline, to start very small. Have two weeks or one month, set expectations and then work out of that, because the last thing you want is to spend all these resources, not just money, but time from your team."