Zeon Systems Wants AI to Handle the Worst Part of Freight Brokerage

The Macro: Freight Brokerage Runs on Phone Calls and Stress

I’ve spent time watching freight brokers work, and it’s one of those jobs that looks simple from the outside and is absolutely brutal on the inside. A broker’s day is a loop: find a carrier, negotiate a rate, book the load, then spend the rest of the day checking in on that shipment while simultaneously trying to book the next one. Multiply that by 15-30 active shipments per broker, and you start to understand why turnover in freight brokerage is brutal.

The transportation management system (TMS) market is big. Estimates put it somewhere around $14 billion globally and growing. But most TMS platforms on the market today are essentially databases with dashboards. They organize information. They don’t do work. McLeod Software, MercuryGate, and Turvo are the names brokers know. They’re solid tools for record-keeping and workflow management, but the actual job of calling carriers, negotiating rates, tracking shipments, and handling exceptions when something goes wrong? That’s still a human on a phone.

Here’s the part that matters: mid-market shippers (the companies that actually need to move goods but aren’t big enough to have dedicated logistics teams) get squeezed. The big freight brokerages like C.H. Robinson, XPO, and Echo Global Logistics have armies of brokers and proprietary tech. A shipper doing $10 million in annual freight spend doesn’t get that same level of service. They get whoever picks up the phone.

AI voice technology has improved to the point where it can handle structured conversations with real humans. Customer service bots are table stakes at this point. But applying that same technology to freight, where conversations involve negotiating rates, confirming pickup windows, and resolving exceptions like “the driver showed up but the dock is full,” is a harder problem. The conversations are less scripted. The stakes per call are higher. And the other party on the line, the carrier dispatcher, is usually experienced enough to tell when something feels off.

The Micro: An AI That Actually Picks Up the Phone

Zeon Systems came out of Y Combinator’s Spring 2025 batch with a straightforward pitch: build a TMS where the AI doesn’t just organize your freight data but actually does the work of a broker. Their AI Transportation Coordinator calls carriers, tracks shipments in real time, and handles exceptions as they come up. The claim is that it makes brokers 2-3x more productive.

The founding team brings an interesting mix. Bronte Kolar is the CEO and co-founder. She built electronic systems for electric aircraft and high-density batteries before moving into computational biology research and working as an early engineer at LatchBio. That’s a non-obvious background for logistics software, but it signals someone who’s comfortable building systems that interface with messy, real-world operations. Tahir D’Mello, the CTO and co-founder, spent his career building software and ML for science labs, with computational research stints at IIT and Yale plus biotech and pharma roles across India and the US.

Neither founder came from trucking, which is either a red flag or a feature depending on your perspective. The argument for outsiders is that they’re not constrained by “this is how it’s always been done” thinking. The argument against is that logistics is full of edge cases that only people who’ve lived through them truly understand. I lean toward the outsider advantage here because the core technical challenge is AI voice and real-time coordination, not domain knowledge that can’t be hired for.

The product itself lives behind a landing page right now, which is typical for a company at this stage. You’re not going to find screenshots or a self-serve demo. That’s fine. The sales motion in freight software is always high-touch. Nobody buys a TMS from a website.

What I find compelling is the positioning toward mid-market shippers. The Fortune 500 companies have already built or bought their way into sophisticated logistics operations. Going after the companies one tier down, the ones spending $5-50 million on freight annually, is a smart wedge because those buyers have enterprise-level needs but mid-market budgets. If the AI can genuinely replace or augment broker headcount, the ROI math gets very simple for those buyers.

The Verdict

The freight brokerage market has been waiting for this kind of product. Previous attempts at automation in logistics focused on optimization (better routing, better pricing) rather than execution (actually making the calls and handling the exceptions). Zeon is going after execution, which is where the labor cost sits.

At 30 days, I’d want to hear recordings of the AI handling real carrier calls. The pitch is only as good as the voice agent’s ability to navigate an unscripted conversation with a trucking dispatcher who’s juggling 40 loads of their own.

At 60 days, the question is retention. If brokers use the AI for a week and then go back to the phone because “it’s faster,” that tells you something. If they never go back, that tells you something much better.

At 90 days, I’d be watching the competitive response. MercuryGate and Turvo are not going to sit still while an AI-native upstart eats their lunch. The question is whether Zeon can build enough of a moat through model training on real freight conversations that the incumbents can’t just bolt on a voice agent and call it done.

The founding team is technically strong, the market is massive, and the pain point is one that every broker feels every single day. That’s a good foundation.