Sam Basu quit Google in early 2023, right when everyone and their cousin was either freaking out about or getting hyped over ChatGPT. Like many engineers watching the AI boom unfold, he wanted in. He tried launching a few AI ventures, testing ideas, seeing what might stick.
Nothing did. Until a friend called asking for help with customs paperwork.
That phone call led Basu down a rabbit hole that would completely change his trajectory. He started cold-calling customs brokers around Los Angeles, expecting to find at least some semblance of modern technology integration. What he discovered instead was an industry stuck somewhere in the 1990s, complete with fax machines and filing cabinets overflowing with paper.
The real turning point came during a FaceTime call with his first potential customer. She gave him a virtual tour of her office, camera panning across stacks upon stacks of manila folders. Basu immediately booked a flight to see it in person.
When Reality Hits Different
“That was the eye-opening moment. There’s just papers and papers,” Basu told TechCrunch. He was shocked that this is how the backbone of global trade actually operates. Every single physical item you own that wasn’t made domestically passed through this system. Your phone, your clothes, your coffee maker. All of it required someone to manually process mountains of paperwork.
It’s the kind of thing you never think about until you see it with your own eyes.
Basu teamed up with Arushi Vashist, another ex-big tech engineer who’d been working at LinkedIn. Together they built Amari AI, and the startup has already racked up over 30 customers who’ve collectively moved more than $15 billion worth of goods using their platform. That’s not small potatoes for a company that just came out of stealth mode.
First Round Capital and Pear VC saw enough potential to co-lead a $4.5 million funding round. Todd Jackson from First Round credits Basu’s old-school hustle, the kind where you actually show up to trade conferences and talk to people face-to-face instead of just spamming LinkedIn messages.
The Perfect Storm for AI Adoption
Here’s where things get interesting. The customs brokerage industry is facing a perfect storm of challenges that make it ripe for AI disruption. The licensing exam has a pass rate hovering between 10% to 20%. By law, customs brokers must be U.S.-based, so companies can’t just offshore the work to cut costs. And experienced workers are burning out and leaving the field.
Enter Trump’s chaotic trade policy, which has turned customs brokers into some of the most important people in business right now. Chris Bachinski, CEO of 125-year-old firm GHY International, told TechCrunch that many importers don’t even have their own compliance teams. They’re completely dependent on brokers to figure out how sudden tariff changes affect their shipments, especially goods already in transit.
Bachinski became one of Amari’s early adopters after catching one of Basu’s presentations at a trade show. GHY isn’t a mom-and-pop shop, but it’s also not FedEx. They’re in that middle ground where staying competitive means embracing new tech before everyone else does.
The biggest pushback? Employees worried about losing their jobs. Bachinski’s message to them has been simple: don’t worry. He sees AI as a growth tool, not a replacement tool. The mundane data entry gets automated, freeing up humans to focus on the relationship and compliance work that actually requires expertise and judgment.
How It Actually Works
Amari’s AI agents constantly monitor trade rules and update their reasoning whenever regulations change. Before this kind of automation, sudden policy shifts meant brokers had to manually research new rules, slowing down their ability to clear cargo. In an industry where time literally equals money, those delays add up fast.
The company has trained its models on over a million documents related to shipments it’s already processed. Basu notes they’ve been using off-the-shelf models so far, but they’re building toward their own specialized systems. Some customers opt out of having their data used for training, and Amari anonymizes everything before feeding it to the models anyway.
“We do not sell their data, and we make sure that their data is theirs,” Basu emphasized. In an industry dealing with sensitive commercial information, that kind of assurance matters.
It’s worth noting that optical character recognition software has existed for years, but Basu says it’s limited and brittle. The difference with modern AI is the ability to actually understand context, not just digitize text from paper. That’s the gap Amari is trying to fill.
The Bigger Picture
Bachinski made a joke that hits different when you think about it: last year was the first time in history that customs brokers’ families actually understood what they do for a living. Trade policy suddenly became dinner table conversation across America.
That visibility is both a blessing and a curse for the industry. More attention means more pressure to perform, but it also means more opportunities for companies like Amari to step in and provide solutions. The customs brokerage world is waking up to the fact that technology is going to transform their industry faster than most players realize.
Whether Amari becomes the dominant player in this space remains to be seen. But Basu’s willingness to actually understand the industry from the inside out, to show up in person and see the manila folders for himself, gives the company a better shot than most AI startups chasing problems that don’t really exist.
Sometimes the best business ideas aren’t found in Silicon Valley pitch decks, but in dusty offices filled with paperwork that nobody outside the industry even knows exists.
