A team of former Big Tech engineers is taking aim at one of the most antiquated corners of global commerce. Amari AI just emerged from stealth with AI-powered software that helps customs brokers navigate the labyrinth of shifting trade policies - a problem that's gotten dramatically worse under Trump's tariff whiplash. The startup's agentic AI platform automates compliance workflows that currently rely on spreadsheets and manual research, potentially saving brokers thousands of hours per year while reducing costly errors.
The customs brokerage industry has a problem that's gotten exponentially worse since Trump returned to office. Every time a new tariff drops or a trade rule shifts - which happens almost weekly now - thousands of brokers scramble to manually update their compliance processes. Amari AI thinks it has the answer.
The startup, founded by engineers who previously built systems at major tech companies, is rolling out custom AI-powered software that automates the grunt work of customs compliance. Instead of brokers spending hours digging through Federal Register updates and cross-referencing HTS codes, Amari's agentic AI does it for them - parsing policy changes, flagging affected shipments, and suggesting compliance strategies in real-time.
"Customs brokers are drowning in regulatory complexity," according to TechCrunch's exclusive coverage. The timing couldn't be more critical. Trump's on-again, off-again tariffs on Chinese imports, sudden Section 232 investigations, and unpredictable trade negotiations have turned what was already a complex job into a daily crisis management exercise.
The market opportunity is massive. The customs brokerage industry handles over $3 trillion in U.S. imports annually, yet most firms still operate on legacy software from the 1990s or cobbled-together Excel workflows. Amari's approach is to build custom AI agents for each broker's specific needs - one might specialize in automotive parts classification, another in pharmaceutical compliance.
What makes Amari's tech interesting is the agentic AI architecture. Rather than just surfacing information, the system can actually take actions - automatically reclassifying products when tariff codes change, generating protest filings when duties seem incorrect, or routing urgent compliance issues to human experts. It's the kind of workflow automation that has been teasing with GPT-4's function calling, but applied to a vertical that desperately needs it.
