The people building your freight tech should have lived the work it's meant to support. They need to have sat in your seat. The most important qualification a software provider can have isn't on the spec sheet; it's whether they've ever had to cover a load when the market moved against them.

Freight has a particular kind of urgency to it. A rate that was right an hour ago can be wrong now. A lane that looks clean in a spreadsheet behaves differently when the driver, dispatcher, appointment and weather all get a vote. You feel that pressure in your body when you're the one responsible for covering the freight. You don't feel it when you're three layers removed, looking at aggregated market data.

Brilliant data scientists and economists can build sophisticated models off macroeconomic indicators and miss completely, because they never talked to a driver or dispatcher or sat through a negotiation. The math might have been elegant, but it just didn't match with how freight actually moves. When they stop staring at outside factors and start looking at the behaviors that actually drive a buy, everything changes. The accuracy improved because the people behind it finally understood the job.

End-to-end operational experience is the difference between software that fits your workflow and software that fights it. A button might be placed somewhere that makes perfect sense to an engineer, and zero sense to the person whose job it is to click it all day. You end up with an expensive tool that everyone quietly routes around for years, only to discover your mistake when a contract renewal is up for discussion.

If your tech provider needs you to explain what a tender is, or why margin visibility across an RFP lifecycle matters, or what makes a lane defensible to a shipper, that gap is going to show up in the roadmap, the support tickets and the features that look impressive in a demo, but collapse the moment they meet a real book of business.

When a shipper says they got burned on a backhaul, or that their out-time doesn't line up with their in-time, or that a rep got too aggressive on a lane they never buy well, there should be no learning curve in the conversation. You should be on high alert if your provider is constantly telling you they’ll “look into it and get back to you,” as that might mean they’ll ask their large language model of choice after the call.

We live in a moment where artificial intelligence content is outpacing authenticity and experience. It has never been easier to sound like an expert. Anyone can generate a confident-looking white paper, polished pitch or feature list replete with the right vocabulary. Pricing AI might generate a number on a niche commodity that it’s never seen, borrowing instead from data that didn't apply. No footnote or hesitation — just a wrong answer delivered with total conviction.

This dynamic is everywhere, not just in the models but in the marketing around them. The surface area of plausible-sounding-but-hollow has exploded, while the cost of choosing wrong has gone up. Investing in the right freight tech today carries a higher price than it used to, because the noise is louder and the imitations more convincing.

Lived experience is the one thing AI can't fake. So when you're evaluating freight technology, whatever the category, get underneath the demo. Ask who built it and what they did before they built it. Ask whether their team has ever covered a load, sat with a dispatcher, or lost margin on a lane they misjudged. Don’t skip over the introductions or let your eyes wander to another screen. Notice how the first conversation feels: Are you explaining your business to them, or are they finishing your sentences? 

The right answer will come from the provider who's been on your side of the table — who can tell you, plainly and without bluffing, when their tool should be trusted and when your judgment should override it. In an industry running this fast, that's the whole game.

Carly Gunby is vice president of revenue at Transfix.