How to Distinguish AI Fact vs. AI Fiction in Freight Tech

Artificial Intelligence (AI) is all the rage in Freight Tech in 2025. Promises of fully automated workflows, 100x ROIs, and "absolute gamechangers" abound.

We all know that AI is going to change how business is conducted. Everyone's chasing AI's competitive edge, but only some of the promises made by Freight Tech AI vendors will stand the test of time. Being knowledgeable about the fact and fiction of AI in Freight Tech helps you avoid costly pitfalls and get the most long-term value out of this new, transformative technology.

In this article, I'll be reviewing a few common areas to think about when trying to distinguish fact from fiction when it comes to AI in Freight Tech. I'm going to focus more on Track & Trace topics for Brokers, but I'll keep the ideas generalizable to different categories of software in Freight Tech. Lastly, I use "AI" in quotes because most of what is offered in the market as AI is actually rule-based automation; what is AI is mostly a wrapper on a standard LLM or voice AI provider.

The truth is, no process can be completely automated, simply because there are too many breaking points in any core freight workflow. Humans need to be involved, approve information, and often enter data. Then, when it comes to something like tracking, outside parties and their systems are involved, which means even more breaking points.

A better framework for thinking about this is the automation of discrete actions (a check call before a pickup appointment to confirm location and ETA). These may be automated in ~40% of cases (and this number is rising over time). Why that number isn't 100% is evident to anyone who has actually worked with drivers... they don't always answer and aren't always helpful. Even with automated sequences in place, you should expect humans to intervene in over 50% of cases.

That is just one discrete element of work... each person does many of these, some of which are more or less automatable. You will be able to accumulate more and more of these discrete automated actions over time, but humans will still be the connective tissue.

Long story short, you'll still need people for a while to handle the work if your "AI" fails. Oh, and by the way, the humans are going to need to hold the barrage of responses that don't quite get the answer you're looking for.

Anyone who has received AI-generated content from AI SDRs, recruiters, or even colleagues knows that it can be a confusing experience. A recent HBR case study found that 32% of people who received AI-generated work slop were less likely to work with the sender again¹. Given that Brokers operate in a relationship-based business, sending AI messages recklessly can harm credibility with carriers, customers, and drivers.

The most skillful Brokers will implement automation and AI surgically. They will carve off low-value work, automate as much as possible, allow people to retrain to manage this new workflow, then repeat. Candidly, they should never expose their most valuable customers and carriers to excessive automation. Excellent service and a human touch will always be a differentiator in this industry, and that holds true even through the AI revolution

While there are individual cases where you can truly plug and play, most of the value lies in molding your operation to complement your automation and AI initiative. The organizations that best marry great AI tools with operations that support exceptions, continuous feedback, and flexibility in day-to-day work will have the most significant advantage.

According to Satya Nadella, the slowest part of this AI revolution isn't technological advancement; it's getting people to change how they work³. Every Broker has those high performers that like to "do it how they've always done it". While there is nothing wrong with that, incomplete or uneven adoption of solutions leads to problems. Quality change management and communication are key to getting your team on board and helping your operation adopt the technology.

It's ideal if your vendor has opinions or best practices on how your organization should adopt the solution. They should share what's worked well and the pitfalls to avoid. Even better is if they choose to integrate these best practices into their solution and make it more opinionated. While that doesn't work for everyone, driving best practices within the technology makes it easier to get the team on board.

I recommend one best practice here: process map a before-and-after with your AI and automation implementation. This doesn't have to be complex —just a simple before-and-after diagram with the affected individuals as stakeholders. Likely, one change isn't enough; you'll have to revisit this process a few times before you reach a long-term, steady state. That additional effort is worth it; it gets people bought in and motivated, and avoids the pitfall of disengagement and silent non-compliance from team members.

Most of the industry is squarely in the testing phase of AI adoption. Given the promises of AI vendors and the stories circulating, there has been little pushback against the extravagant claims of rising Freight Tech startups. Don't get me wrong—I am a serious optimist about this technology—but we need to be realistic about what can and cannot be automated.

That being said, the wave is coming, and in its early phases, every Broker should be getting their organizational data and people in a position to adopt the technology and start experimenting. That means starting with data cleanup and standardization. Here is a short checklist:

Getting the delicate balance right for using AI in live production workflows is challenging. It might be straightforward for simple, low-hanging, repetitive internal work, but it becomes much harder as your automation volume grows or your process increasingly involves outside parties. You're going to want to configure a lot and provide feedback on who to reach out to and who not to, and how to route all the responses.

This configuration process and its maintenance will never end, but there is no doubt it's worth the effort, given the value it will deliver. The question is, who will manage this maintenance? Is it you or your vendor? If it's you, how complex are the changes, and could things be broken easily? Having someone accountable for the uptime and performance of your automation and AI initiative is definitely a best practice.

Whether we like it or not, the age of AI is upon us. It's not going to solve all of our problems; it's not going to solve none of them; it's somewhere in between.

Hopefully, this article helps you think critically about the realities of AI and Automation as of October 2025. I recommend that decision-makers at Brokers begin preparing their data and teams for the change and start experimenting with the discrete automation of low-value work. The good news is that this transition will take time, and much of what makes companies differentiated already (excellent service and relationships) will still apply in the future.

I welcome dissenting or agreeing opinions! Share your thoughts below.

I'm Taylor Diem, the CTO of TrackFlo (a Freight Tech company focused on AI paired with human workflows in Track & Trace for Brokers). I spent half a decade at FourKites helping our team and customers extract value from data. Previously, I worked in advanced analytics at Deloitte Consulting, and I graduated from the Wharton School of the University of Pennsylvania with a concentration in Operations & Information Management.