When the truck talks back

Panelists said the key question is how far fleets are willing to let AI reshape their culture and processes. (Karen Foote/American Trucking Associations)

PITTSBURGH — The driver’s phone buzzes and sends an audible alert. He glances down by instinct and starts to lift the handset, but before the notification can be read, a calm voice in the cab cuts in: “Please keep your eyes on the road.” With 80,000 pounds under the driver’s control, the phone goes back down. The risky behavior is corrected in the moment, before it can lead to a crash.

Scenes like this are becoming more common as fleets shift from reactive safety programs to proactive, artificial intelligence-powered driver coaching. Fleet executives and technology providers discussed how this trend is reshaping fleet safety, driver performance and overall carrier operations during The AI Summit, hosted Sept. 22-23 by American Trucking Associations’ Technology & Maintenance Council.

Panelists said the key question is how far fleets are willing to let AI reshape their culture and processes.

Real-time coaching

Dan Lambert, senior director of product management for Lytx, said AI-driven in-cab coaching systems use machine vision and telematics to interpret conditions around the vehicle and communicate with the driver immediately, marking a shift from safety programs that have historically been reactive.

“You can just tell, real-time, to the driver, ‘Hey, you know, why don’t you put your phone down,’” Lambert said. “We found in our data — when you turn on the in-cab alerts — over 90 percent of the behaviors just go away overnight.”

Shifting toward continuous, in-context feedback and away from visits to the safety office is improving in-cab behavior, said Adam Kahn, chief business development officer for Netradyne. He stressed that the timing of coaching is as important as the message.

“If I push that coaching event to the end of the day, I am still two times more effective coaching in the cab,” he said. “If I push it to the end of the week, I am 10 times more effective coaching in the cab than a week later.”

Dan Benacquista, vice president of strategic operations and planning at Modern Transportation, said the carrier focuses heavily on “precursor behaviors” such as distraction and near misses and has seen a clear difference.

Image

Benacquista. (Karen Foote/American Trucking Associations)

“If you can conduct that coaching as close to when the behavior happens, it is going to be more effective,” he said. “After-the-fact coaching helps, but it is no substitute for prevention.”

Dissecting the data

Fleet safety teams can face paralysis by analysis when reviewing data from electronic logging systems, telematics systems, engine diagnostics and dash cameras. The sheer volume of alerts and video clips can make it difficult to identify the events that matter most. But tech providers said AI agents that continuously scan incoming data can surface trends a human analyst might miss.

“You can have a team of agents watching and monitoring in real time all the data that is flowing in,” Lambert said. “The real key there is then also bringing the human back in to decide what to do with that data.”

Benacquista said AI helps his team “distill large amounts of data into insightful actions” and “cut through the clutter,” provided fleets remain engaged and refine the prompts and guidance they give the tools.

Kahn noted AI can also connect safety outcomes to operational bottlenecks elsewhere in a fleet’s network.

“Think about a delay at the distribution center,” he said. “If there is a delay at the distribution center, what does that do to your driver? Does that delay get adjusted in their workflow, or is the delay now the burden of the driver to make up in an unsafe manner?”

By correlating loading delays, route changes and safety events, Kahn said fleets can identify when operational pressures are pushing drivers toward higher-risk behavior.

Video moves from shield to scalpel

J.S. Bouchard, co-founder and chief product officer at technology provider Isaac Instruments, said outward-facing cameras can help exonerate drivers after crashes and protect fleets from potentially costly courtroom visits. But as AI models have improved, video has become a more precise safety tool.

Image

Bouchard. (Karen Foote/American Trucking Associations)

“What we do not want is a video of my driver with a phone in his hand. That is a liability, which I do not want,” he said. He stressed that customers not only want to know when a risky behavior occurs, they also want systems that reward quick self-correction.

“If the driver has corrected the behavior within the time that I have allowed him, that is good,” Bouchard noted. “The solutions have really evolved. It is the awareness and it is the change of behavior before the event is actually recorded.”

Context matters

That approach also can ease driver concerns about privacy and “gotcha” safety cultures. The systems continue to monitor behavior, but if a driver responds quickly to an alert, no clip is generated and the back office never sees the event.

AI also provides context for traditional telematics metrics. Kahn recalled that years ago, hard-braking events could get a driver fired regardless of context. But by reviewing video and telematics together, fleets now can assess causality and severity.

“It could be [the driver] did hard braking because someone was jaywalking,” he said. “That is a really good decision to do a hard braking, versus my eyes were diverted and I did a hard braking because I was distracted.”

Those types of examples allow fleets to use data to define and reinforce positive performance, Kahn stressed. With tens of thousands of minutes of driving recorded for each driver every month, fleets have a large sample of safe behavior they can capture, quantify and recognize.

“Having data actually supports how you model what good looks like, versus chasing what bad looks like in your organization,” he said.