PS Logistics, a full-service supply chain provider with a family of regional carriers under its umbrella, wanted its first AI application to serve drivers. The company was recognized as a CCJ Innovator this year for its AI voice agent that it co-created with a tech company to enable its drivers across roughly 17 operating companies to call a single phone number and retrieve details on things like loads, truck repairs and more.
“It doesn't sound as impressive now because it's a bunch of voice agents that you can now purchase, but that's how we started our AI journey,” said PS Logistics Vice President of IT Mauricio Paredes during a panel held last week at McLeod UC26 in Nashville. “A lot of those transactional conversations are happening on the voice agent itself, and that was our first use case: very successful, roughly 70-75% resolution rate on the AI calls. So we took that success and started building additional use cases on top of voice.”
Meanwhile, on a different playing field, Axle Logistics took a different approach to implementing AI.
Rather than pinpointing its AI use case by starting with a specific group, the freight brokerage looked across operations for repetitive tasks. For example, one back-office process that was identified involved opening carrier emails, uploading the documents within and moving information into the TMS manually.
Paredes and Axle’s President Shawn McLeod shared during the panel how their respective companies arrived at different AI strategies and how those strategies have shifted from experimentation to implementation.
From pilot to productivity
Almost every member of the audience in attendance at the panel raised their hand when asked if their company is using AI in some way. But Paredes and McLeod illustrated that it’s not just about using AI but using it effectively.
AI has evolved beyond a technology project for these companies to more of an operational model, though McLeod admitted that Axle has been much slower to adopt.
“We're just on the cusp,” he said, while also noting that the company’s current use of AI has already resulted in reduced employee headcount and an uptick in load count and revenue per head because it has enabled them to take on more customers with less workload.
That has allowed the company to employ a dedicated person to look at audit logs and identify other opportunities to deploy AI.
PS Logistics, however, is barreling down the tracks on the AI train — or maybe more appropriately barreling down the road on the AI convoy.
The company built on its AI voice agent and created a terminal version that all employees can use. Instead of using a person to gauge new ways to use AI, the company’s AI asks users if they want to save whatever function they’re working on as a skill if it recognizes the task is a repetitive action.
“The way that we're surfacing those use cases is we're looking at what type of skills are being commissioned,” Paredes said. “Our particular harness keeps versioning, so if you hit a version like 12 of a particular skill, then we start looking at it to understand what it does.”
He added that another way the company surfaces use cases is by connected related skills together to create what he calls a workflow.
“Everything that the agent is learning eventually gets moved over to be accessible to our drivers via voice,” he said. “Of course, not every workflow is driver facing, but now our users via using the AI agent are actually teaching the AI agent enough business rules and enough business workflows that we can now make them available to the drivers to do some self-service.”
From use case to encouraging use
The biggest hurdle both panelists highlighted in implementing AI was employee buy in.
Paredes said the win from the drivers using the AI voice agent gave the company the jumpstart it needed to drive AI forward, but he noted that only roughly 30% of PS Logistics employees are using the company’s AI. It’s gaining traction, though, as word-of-mouth spreads about what it can do, he added.
McLeod said the grapevine was also Axle’s best tool in acquiring users.
He said the first uphill battle with AI was when the company automated rating, where an AI bot submits a rate on a bid without the employee touching it. It was a trust issue, he said, adding that some of the sales representatives in the beginning found that the AI was leaving a lot of opportunity on the table. But once the AI had time to adjust, Axle saw an increase in volume and margin percent.
“They needed to see success,” McLeod said. “… The more people saw that, the more people would question, ‘Hey, what's going on here? How can we become part of that?’”
He noted that it’s important to get employee feedback on both the pros and cons of AI along the way.
That’s also part of PS Logistics’ process, which Paredes said is now a more formalized approach in which the company marries the capabilities of the agent to the process it’s affecting via valuations and Full-Time Equivalent assessments that compare the AI’s productivity to human workers.
That process gains buy-in as well, but he added that latecomers have benefitted from the work early adopters put in to train the AI, which is what has fueled word-of-mouth adoption.
Those successes, Paredes said, come from continual encouragement of experimentation.
“This is a particular technology that is different from all the other technologies; It actually pays back any exploration with productivity increases,” he said.
From productivity to people
For McLeod, determining use cases for AI was Initially about efficiency and productivity gains, but going forward, he said it will be about giving employees time back to focus on customer relationships and work-life balance.
“There's been a shift in my mentality over the last year that we need to give (employees) more time back, but it's not really even time freed up to do more sales or do more account development; it's more … let's try to still continue to have fun.”
He added that AI can create as many operational problems as it solves if companies automate interactions without considering what happens on the other side of the transaction.
“I don't want an AI bot to call some of the carriers in here and drive them crazy or their drivers crazy with updates every 15 minutes,” he said, adding “I don't need your bot calling me about all 6,000 loads I have on the board available today because it just bogs everyone down, and then it becomes an inefficient game.”
Paredes said PS Logistics is also protecting relationships from AI.
The same voice agent that drivers can call if they choose to interact with the AI could be used by the driver managers to automate calls to drivers, but PS Logistics prohibits it, alongside the other AI governance guardrails in place.
“Nobody wants to feel like they're being managed by a machine. That's not what this is,” Paredes said.
But both PS Logistics and Axle use AI to help employees develop better communication skills to improve their relationships with drivers and customers.
Paredes said PS Logistics is working to develop AI that helps the driver manager have a more effective call with the driver rather than have the AI perform the call. An example he offered is the AI putting all the data surrounding any given driver into context and providing a list of things the manager needs to touch on before the call ends.
Axle is using AI for employee training in a different way. The company has an AI tool that provides 10 different personalities for employees to practice cold calling. The tool scores them on their tone of voice and energy as well as on what questions they asked.
“The (AI) is there to help them have more time to get on the phones,” McLeod said.
