5 Takeaways From Bison Transport and KBX Logistics on What Running AI in Production Looks Like Inside a Freight Operation

Artificial intelligence is one of the loudest and noisiest topics in trucking right now. While AI seems to be top of mind for everyone, it’s hard to find operators who have actually implemented it to support their networks.

That’s what made the fourth session of the TCA’s 2026 Online Leadership Series so valuable. Hosted by the Truckload Carriers Association, the session included Optimal Dynamics’ Daniel Powell as moderator plus two executives who have successfully embedded “Everyday” and “Transformational” AI into their operations:

• Tracy Reiss, SVP of Transportation for KBX Logistics
• Brad Gerrard, CIO of Bison Transport

Here are 5 key takeaways from their discussion.

1. Waiting Isn’t the Same as Doing Nothing

Build vs. buy has always been a difficult decision. It’s getting even harder now as trucking companies have to choose between building, buying, or waiting. Brad explained that the economics keep moving underneath you. He pointed to the “tokenomics” of large language models as a variable that didn’t exist previously.

“A decision made six months ago can be economically wrong today, just based on how those costs are being passed through.”

Waiting is a defensible position, but it’s important to wait on the right things:

“Waiting isn’t the same as doing nothing,” Brad said. “The mistake is waiting on the wrong thing. Fixating on that one use case and saying, this is the one I’m going to get. Don’t wait on leveraging everyday AI ... spend the time making sure your data and your process foundation are in order. That work doesn’t get cheaper or faster by delaying it.”

Tracy made the same point from an operations perspective. If you have multiple operating systems holding critical data, integrate them and make that data available to whatever consumes it. Without that solid data foundation, transformational AI is “pretty tough.” Siloed data behind big walls means watching everyone else capture the value.

‍

2. Your Data Architecture Sets Your Ceiling

Bison made the decision years ago to move to event-driven architecture. That move looks prescient now. In practice, it looks like this: When something happens here, get it over there, in the format it’s needed, as fast as the transaction happens.

The result is speed in operations.

“We had an order show up, we had a tender show up — time to plan it,” Brad said. “It’s not an hour, it’s not two hours. I think it’s sub-five minutes, sub-minute in a lot of cases here now. That’s the speed that AI likes to work at.”

Most carriers are held back by legacy systems that want to work in overnight batches, and a decision layer that can only move as fast as the data reaches it. Event-driven architecture is what let Bison start decoupling from that. Now, the order lifecycle and the decisions move at the same pace, while Bison contains the collateral effects on the systems that still run in batches.

Daniel drew the comparison to businesses sitting on capital that can be available whenever opportunity appears. Data now works the same way. If you know where your data lives, who controls it, and how fast you could get to it, you can choose a tool and move forward with implementation at any time.

3. Call It a ‘Recommendation Engine’

As the discussion moved to the concept of “transformational AI,” both Optimal Dynamics customers shared their views on how automation is removing inefficiencies. They noted that it’s important for employees to know they can make the final decision to accept the optimized recommendation presented to them.

Asked what she’d do differently, Tracy said she’d change one word:

“I had a conversation with a carrier this morning about Optimal Dynamics, and I said, if I were to go back and do it all over again, my biggest piece of advice was: we talked about Optimal Dynamics like a decision engine, and we needed to talk about it like a recommendation engine.”

Reframing the platform doesn’t remove accountability. Rather, it enhances accountability for the KBX team.

“We needed to ensure that we understood from our people what would cause them to not take the recommendation,” Tracy said, “but also that they’re accountable for the decisions that are made.”

‍

4. Planners Don’t Disappear

Every carrier asks what happens to the planning team, and the honest answer is not “nothing changes.” Brad’s guidance was to have that conversation directly and early. He emphasized the importance of direct conversations over vague reassurances.

“You really have to pivot people to think in a more senior capacity,” he said. “Your role is changing, the expectations are higher on you. Why? Because you need to own the exceptions. You have to validate the patterns.”

Day to day, planners are now asked to be more strategic and analytical, answering questions like:

• Is a seasonal shift moving the freight network?
• Is there a disruption building somewhere in it?
• What doesn’t look right today?

“It’s less about that task-level clicking and assigning and moving things around,” Brad said. “It’s to really think about managing the network and managing the freight patterns within that network.”

The underlying logic is about where human judgment actually holds its edge. Institutional knowledge is enormously valuable and scales well when operating a small fleet. It stops scaling well in the face of network-wide trade-offs across thousands of trucks.

Tracy’s team talked at length about modern architecture and data foundations, and not nearly enough about what it meant for the people doing the job. AI is scary to a lot of people right now, but being open and transparent early is better than repairing trust later.

‍

5. The Adoption Surprise

Tracy went into the implementation process expecting adoption to be a challenge. Her planners had always picked the freight, and she feared the “recommendation” framing would lead to more overrides. That was not the case.

“I underestimated, I’m sad to say and proud to say at the same time,” she said, “because our people, once reframing it, they wouldn’t go back.”

Once planners saw the system as something helping them make better decisions, ownership went up rather than down.

“It can run 20,000 scenarios; they can’t,” Tracy said. “But it can present alternatives to them to say, hey, this is the best, this is what we recommend, here’s some alternatives, here’s the trade-offs.”

When you include people in the decision and show the trade-offs, adoption takes care of itself.

‍

Next Steps: Getting Started With AI

No matter where you stand in evaluating, selecting, and implementing AI tools in your operations, there are actions you can take right now to streamline the process.

Each week, our team speaks with fleets that are at different points in that process. We can provide guidance and make recommendations on what to do next to maximize your fleet’s performance and profitability.

‍Get in touch to start the conversation around artificial intelligence for your operations.

Frequently Asked Questions

What is the difference between everyday AI and transformational AI in trucking?

Everyday AI covers today’s widely available tools, like language models and assistants that speed up individual and team-level work. The barrier to entry is low, and Brad Gerrard advised to start leveraging it now. Transformational AI operates at the network level, automating and optimizing decisions like load planning, freight allocation, and dispatch across thousands of trucks. Transformational AI relies on a clean, integrated data foundation. Tracy Reiss suggested that transformational AI is difficult without that foundation in place.

What happens to freight planners when AI is implemented?

Planners spend less time on task-level clicking, assigning, and moving loads around. They spend more time owning exceptions, validating patterns, and reading the network. Tracy Reiss expected pushback from her planners at KBX but received the opposite. Once her team saw the platform as something that helps them make better decisions, their level of ownership increased.

What is the biggest mistake carriers make when implementing AI?

Waiting on the wrong thing. Brad Gerrard said that fixating on a single use case or holding out for the perfect one wastes time that could be spent on data quality and processes. The data foundation work doesn’t get cheaper or faster by delaying it. Tracy Reiss shared a mistake around language. KBX introduced the platform internally as a decision engine, when it should have been called a recommendation engine. That reframing helped planners better understand their accountability.

What are some results carriers are seeing by implementing AI in their operations?

Results vary by network and by adoption rate. Learn more about the results carriers have achieved by exploring our case studies.