Synopsis: FlavorCloud CEO Rathna Sharad dive into how artificial intelligence (AI) is being used to optimize international trade in a way that mitigates the impact of rapidly fluctuating tariffs.
Cross-border commerce sounds glamorous until you hit the maze of tariffs, shipping rules and ever-shifting holiday calendars that differ in 220-plus countries. Rathna Sharad discusses how artificial intelligence can keep merchants from drowning in that complexity. Sharad, whose team builds software for international shipping and compliance, says the answer is “domain-specific” AI that understands trade codes, carrier rules and local laws well enough to do in seconds what human brokers once did with 800-page reference books.
Her group’s first target was product classification. A women’s cotton dress, for example, carries a six-digit Harmonized System code recognized worldwide, but every country tacks on extra digits that dictate duties and inspections. A custom model called “Flash AI” now assigns those codes automatically, then hands the parcel to another model that guarantees landed costs—duties, taxes, fees—before the shopper clicks “buy.” Some agents rely on classic machine-learning, others on retrieval-augmented generation, yet all feed a single orchestration layer that decides whose output to trust.
The payoff isn’t just lower cloud spend; it’s resilience. Tariff tables can change overnight, and a misclassified shipment means someone—usually the merchant—eats the surprise bill at the border. By tying models to real-time data and version-controlled rules, Sharad’s platform rolls out updates the moment new rates become law. She sees the same approach replacing manual customs brokerage and enabling dynamic sourcing, where a brand can switch factories or fulfillment centers on a week’s notice without rebuilding its tech stack.
Still, she cautions, AI is not a silver bullet. Guardrails like peer review and sandbox testing are critical when an errant prompt could stall a container or expose customer data. Her advice: start small, measure the minutes and dollars saved, and invest in upskilling your supply-chain team so they can question the models they deploy. ROI—not novelty—should decide what goes into production.
