How Spartan’s Operators Became AI Agent Builders

Spartan’s Operators Went From AI Skeptics to Agent Builders

When Spartan Carrier Group began exploring artificial intelligence about two years ago, the idea drew skepticism from the company’s operations staff. Today, the Fort Worth, Texas-based carrier is using an AI agent to automate customer reports and is preparing to test another workflow involving driver payroll.

Kendrick Dawson, Spartan’s vice president of operations, was among those who initially doubted the technology. Dawson joined the carrier from Ryder in 2021 and said the team was not convinced when founder and CEO Carlos M. Llanes Jr. began discussing AI.

“At the time, all of us looked at him like he was absolutely insane,” Dawson told FreightWaves. “We’re like, bro, it’s never going to happen. What are you talking about? Where are you getting this information? Two years later, here we are.”

Spartan operates about 150 trucks and 400 trailers with roughly 170 drivers. The company hauls just-in-time automotive freight, including freight for Toyota, its largest customer. Spartan began hauling directly for Toyota in 2023 and now handles freight at Toyota plants in San Antonio and Blue Springs, Mississippi. It also works with a group at the Mazda Toyota Manufacturing plant in Huntsville, Alabama.

The carrier grew from a single truck to its current size in about five years. Dawson said Spartan wants to continue expanding without simply adding more dispatchers. The company runs around the clock with four dispatchers: one assigned to each shift, along with a weekend and relief dispatcher.

“We don’t want to let anybody go, we don’t want to fire anyone. That’s not the point,” Dawson said. “It’s how can I maximize my operation with the same headcount that I have now, and allow those people an opportunity to grow and learn to do different things.”

Automating a repetitive dispatch task

The company’s first AI agent was designed to handle hourly customer tracking reports. Some Spartan customers require updates in specific formats and on fixed schedules. Before the agent was put in place, dispatchers created the reports manually in Excel.

Preparing one report generally took seven to 10 minutes and could take as long as 15 minutes for a newer employee. The reports had to be produced repeatedly while dispatchers were also monitoring freight, communicating with drivers and solving problems across the network.

When Dawson asked dispatchers which task they would eliminate, the hourly route updates were the consistent answer. The work was necessary, but it took time away from operational decisions.

Spartan built the agent through Foundry, an AI platform launched by transportation management system provider Alvys in August. Foundry allows carriers to configure agents within the TMS, using the operational data already stored there.

Dawson met with an Alvys staff member once to learn how to use Foundry and met again to provide the report requirements. A sample was returned in about two days, and the agent was operating within a week and a half.

Dawson said he had forgotten that he had activated the workflow until messages began arriving during the operations team’s 9:30 a.m. call.

“Literally 10 o’clock, 11 o’clock, 12 o’clock, 1 o’clock, perfectly on time,” he said.

The reports matched the dispatchers’ versions “almost 100%,” according to Dawson. He estimated that the first agent returned about 17% of a dispatcher’s workday. The agent can pull live estimated arrival times, arrival and departure information, and last-known locations from Alvys and distribute the information in each customer’s required format and cadence.

Automating one report also gave Spartan a way to identify how dispatchers were spending their time and which types of work interested them. Dawson said the goal is to use the time saved for problem-solving and other responsibilities, rather than to eliminate positions.

Payroll workflow is the next test

Spartan is developing a second agent to support driver payroll, although the workflow has not yet been tested.

The carrier pays drivers a flat weekly guarantee for working five days. Much of its freight moves through relay operations. A shipment traveling from Detroit to San Antonio, for example, can pass through six solo drivers at handoff points in several states.

Delays can create complications throughout the relay. If one driver runs late, hours-of-service limits may push the schedule back. Spartan may then use an extra-board driver to keep the customer’s freight moving. A driver who loses a day through no fault of their own still must be paid under the company’s guarantee.

Field personnel currently review electronic logging device records from Samsara along with internal notes to determine what happened. Dawson said that process can take much of a Monday.

The proposed workflow would email field teams each day with a list of drivers missing from the dispatch planner. The recipients would indicate whether each case involved a call-off or an unplanned extra-board day. After approval, the system would add the day to the planner and include the appropriate line item in the driver’s settlement report. A weekly summary would list the decisions.

Dawson expects the process could reduce the time spent on payroll by at least half, but the estimate has not been validated because the agent has not yet been tested.

Operators, not programmers

Alvys staff helped build Spartan’s first agent. Dawson plans to edit the payroll workflow himself and wants dispatchers and field operators to learn how to create additional agents.

“I’m not a tech guy, I’m not a coder, I’m not a data scientist. I just run trucks,” he said.

That shift in responsibility is central to Spartan’s approach. Rather than treating AI as a project limited to software specialists, the carrier is asking the people who perform the work to describe the process and help configure the automation.

For professional drivers, the immediate impact is likely to come through back-office tasks such as tracking updates and payroll accuracy. Dawson said payroll is especially important because errors can affect driver trust. The payroll workflow remains a test, but Spartan’s first agent shows how a carrier can use its existing TMS data to automate repetitive work while keeping its operations staff focused on managing freight and supporting drivers.