Why I Work at the Intersection of AI and Global Trade

The question I get most often is: what connects Zentria Flow, Trazeroad, and FixerCV? Why these specific domains?

The answer is always the same: information asymmetry. Every company I've built targets a situation where one party — an importer, a job seeker, a logistics operator — is making an expensive decision with significantly less information than they need. AI is what makes closing that gap possible at the scale and speed required to be genuinely useful.

Why Global Trade Specifically

I started in freight — not as an analyst studying the industry from the outside, but as an operator moving goods across borders. The longer I worked in cross-border logistics, the more obvious one thing became: the cost structure of international trade is almost entirely opaque to the people paying it.

An importer agrees to a product price. They know they'll pay freight. They roughly know they'll pay duties. But the true landed cost — once tariffs, customs fees, port handling, broker fees, currency conversion, and compliance costs are included — is almost never known until after goods have cleared customs and the final invoices arrive.

The gap between what importers think they'll pay and what they actually pay is consistently 20–50% of the product price. That gap costs businesses money on every shipment, distorts profitability calculations, and makes accurate sourcing comparisons essentially impossible.

This is a structural problem in a market that moves trillions of dollars annually. It doesn't exist because businesses are careless — it exists because the information is fragmented across dozens of data sources that have never been properly integrated. Tariff rates live in government databases. Freight benchmarks come from brokers. Port fees appear on invoices. Nobody assembles the complete picture in real time.

Why AI Is the Right Tool for This Problem

Building a system that integrates real-time tariff data from 180+ countries, current freight rate benchmarks, port fee schedules, and customs clearance cost patterns — and delivers an accurate landed cost in seconds — isn't a spreadsheet problem. It requires AI to classify products automatically, match them to tariff schedules, integrate data from multiple sources, and surface a reliable estimate fast enough to actually influence a buying decision before the order is placed.

The same pattern applies to FixerCV: the gap between how a candidate presents their experience and how an ATS scores it is an information asymmetry problem. AI is what makes it possible to bridge that gap at scale — analyzing each specific job description, comparing it against each specific resume, and producing guidance that's actually personalized rather than generic.

Before AI made this tractable, both problems had the same character: the data needed to solve them existed, but synthesizing it was too expensive for anyone except large specialists with dedicated teams. That scarcity — of time, of expert capacity — protected the information asymmetry. AI removes the scarcity, and with it the protection.

The Information Asymmetry Playbook

Every company I've built follows a consistent logic: find a market where one party routinely makes expensive decisions with far less information than they need, identify the specific information gap that causes the damage, and build a system that closes that gap at a price point the disadvantaged party can access.

The failure mode this pattern avoids is building a solution in search of a problem. When you start from an information asymmetry that has observable, quantifiable economic consequences — importers overpaying by 20–40% per shipment, job seekers failing screening for avoidable formatting reasons — you know before building that the problem is real and expensive. Customers who are already losing money on the status quo are motivated buyers.

This also creates a defensible competitive position. The information gap itself is obvious in retrospect; the hard part is having the domain expertise to know exactly which data sources close it, how to keep that data current, and how to weight conflicting signals accurately. That expertise comes from the operational experience that precedes the product — years running freight before building import cost intelligence, not years studying import cost intelligence from the outside.

Why Global Trade Was Underserved by Technology for So Long

The scale of global trade — roughly $30 trillion annually — is paradoxically part of why the technology serving it lagged behind other industries. The incumbents (freight forwarders, customs brokers, trade banks) built businesses on information scarcity. Their value lay partly in knowing things their clients didn't. Technology that made that information widely accessible was not in their interest to build or adopt.

This dynamic is common in industries with established intermediaries. Travel agencies resisted price-transparent booking platforms. Real estate brokers resisted property databases. The intermediaries who thrived were the ones who learned to add value on top of transparent information rather than in its place.

The same transition is underway in global trade. Tariff data, freight rates, and customs procedures are becoming more transparent — not perfectly, and not quickly, but the direction is clear. The businesses that will do well are the ones who add judgment and synthesis on top of that transparency, not the ones protecting opacity.

What Operational Experience Adds to AI

There's a version of the "AI for trade" thesis that is entirely data-driven: ingest tariff databases, train a model, ship a product. Some companies are building this way, and for structured, well-documented parts of the problem it works reasonably well.

What it misses are the parts of landed cost that don't appear in databases. The surcharges that carriers add in peak seasons that aren't reflected in list prices. The customs examination fees that apply to specific product categories at specific ports based on current enforcement priorities. The documentation errors that trigger holds and generate storage charges. These costs are real and substantial — they appear in every importer's final invoice, often as surprises — but they're learned from operating, not from reading databases.

Running Trazeroad's freight operations gives Zentria Flow a different category of data: the gap between what the databases say should happen and what actually happens on real shipments. That gap is where the most expensive surprises live, and closing it requires operational experience, not just data access.

The Bigger Pattern

Where AI creates the most genuine business value is in information-dense domains where the cost of getting the right information manually is too high for most participants to bear. Global trade is one of the clearest examples: the information exists, but assembling it, keeping it current, and applying it to a specific product on a specific route has historically been too expensive for most importers to do systematically.

AI closes that access gap. That's what Zentria Flow is building: not a better spreadsheet, but a fundamentally different way to access and apply trade cost intelligence. The goal is making the true landed cost of an import as easy to know as the supplier's quoted price — before you place the order, not after it clears customs.

What Comes Next

The intersection of AI and global trade is early. The infrastructure for real-time, comprehensive, accurate trade cost intelligence is still being built. Most importers are still working with manual estimates, outdated tariff data, and reactive cost discovery.

The companies building the right infrastructure now will sit at the center of trillions of dollars of global commerce. That's a worthwhile place to build — and it's where I intend to stay.

For a detailed look at where AI is generating real, measurable results in import cost intelligence today, see AI import cost intelligence. For how AI is transforming supply chain and logistics operations more broadly, see AI in supply chain and logistics.