Photo 238880808 Audioundwerbung |

Using AI to combat data overload

Nov. 10, 2023
Pitstop's recent webinar examined how fleets can apply AI to streamline fleet maintenance and management.

Artificial intelligence, or AI, is playing a growing role in the commercial vehicle industry, whether through AI-enabled dash cams or platforms to monitor tire wear and tear. But on a broader level, AI can now help fleets manage their maintenance beyond a single component.

Jessica Kim, head of marketing and revenue operations at Pitstop, discussed this potential during a Pitstop webinar entitled "AI Revolution in Fleet Maintenance" on Oct. 25 with Ben Roueche, fleet manager for City of West Jordan, Utah. They argued that AI can bolster fleet maintenance in three key ways: By filtering vehicle data, offering predictive insights, and by streamlining fault code comprehension.

The consequences of data overload

According to Pitstop, a fleet maintenance management software provider, the average commercial vehicle makes 15 trips per day, and for a 300-vehicle fleet, that means a fleet manager could be dealing with as many as 7,200 telematics alerts.

Read more: Four ways for fleet managers to avoid data overload

“If we extrapolate that data into a year's view, and let's say that fleet runs without a holiday for 365 days, to analyze all of that information and all those alerts would take a human 570 days,” Kim said.

While that kind of time commitment isn't an option for a busy fleet manager, on the other hand, not accessing these alerts at all can have consequences of their own. Foregoing telematics-generated data leaves fleets to bring in their vehicles for preventative maintenance at certain mile integers, potentially over-maintaining the vehicle and adding to unnecessary downtime and expenses.

“If we take a heavy-duty fleet of 500 vehicles, for example, and based on the average cost of PMs and the average cost of unexpected downtime per quarter, this fleet could be spending anywhere around $130,000 per quarter on PMs alone,” Kim asserted.

Not to mention that without accessing vehicle data, techs and drivers could miss crucial clues that a component is about to fail until the vehicle is at the roadside.

But discovering that a crucial component has given up the ghost after a breakdown is only the first hurdle a fleet may face. Once they get a downed vehicle to the shop, being unable to access and prioritize a truck’s data can put technicians at a disadvantage too, especially with TechForce reporting that diesel technician demand is outpacing supply at 41,369 to 10,989. This can lead to fleets paying higher rates for their repair to retain their workers, as Fullbay noted in its third heavy-duty maintenance report that 75% of shops raised their labor rates, and 92% of those increased their tech pay, too all of which leads to more frequent and expensive shop visits for fleets.

But a fleet manager doesn't have 570 days to parse through telematics alerts, making another solution a necessity.

“AI would do it in 36.5 hours,” Kim offered.

Scheduling PMs with predictive insights

To help fleets cope with data overload, AI platforms can analyze and prioritize fault codes, cataloging them as major or minor issues.

“This allows operators to avoid excess shop visits or part replacements, and instead allocate resources and time to the ones that need it most,” Kim explained.

Then, as the AI platform learns more, it can recategorize faults based on input from the fleet, further streamlining telematics alerts and speeding up diagnostic times. Roueche also noted that having access to these notifications allowed the City of West Jordan's fleet to tailor their service so it caused as little disruptive downtime as possible.

Read more: Six best practices to optimize preventive maintenance

“A lot of times, the best part of it was being able to see stuff that we don't get from the drivers, and have them come in before they have a breakdown or a problem,” the fleet manager said. “We can schedule the vehicles when they weren't using them, instead of having to go and figure out why it stopped on the side of the road.”

Roueche also attested that Pitstop’s AI platform, when paired with their telematics provider, Geotab, helped them deal with an O2 sensor issue quickly and efficiently. Reportedly, the truck came in on a Friday with a check engine light, and thanks to their vehicle data, Roueche's techs changed out the sensors and got the truck back in time for work on Monday.

“That was one of those times where AI showed us exactly what we needed to do,” Roueche recalled. “We knew exactly what we were looking for and got the part that we had to change.”

Streamlining diagnostic codes

 AI-capable fleet management platforms can also clarify diagnostic codes for technicians by using programs such as ChatGPT.

“Pitstop takes ChatGPT and makes descriptive fault codes more digestible and easier to understand, translating them from that robotic terminology that we often see,” Kim noted. “If a driver calls in and asks if there's an issue, or sometimes even without calling in, you can see the alert on Pitstop's dashboard and you're able to quickly take action.”

Granted, a study by Stanford University and UC Berkely found that ChatGPT has grown less effective over time because of how the program learns, but Kim assured webinar attendees that the Pitstop data science team examined ChatGPT with an “extensive testing and fine tuning process” for accuracy and reliability before integrating the program into their dashboard.

Between these three facets of AI, Kim said that fleets could improve shop efficiency by 17%, consolidate their shop visits, and reduce unexpected repairs. Fleets just need to be willing to work with the system to tailor it to their needs.

“It's not a silver bullet,” Roueche said. “The more that you can customize it, the more that it learns about your system, and the more you get to have alerts that actually mean something and aren’t just for a random code from your vehicle.”