Samsara Ride Along pushes fleet safety AI beyond incident flagging

Chief Product Officer Johan Land laid out the differences. Standard camera systems flag discrete events: phone use, close following distance and harsh braking. “They’re typically fairly rare events. It’s a few events a week, of that order of magnitude,” he said.

Ride Along runs for a continuous window and evaluates the full session.

“The difference with Ride Along is that it’s a time period, and then you get the full picture during that, which means that, as I said on stage, most of our pickup is positive,” Land said. “It’s a representative view of the driving, and therefore gives a wholly different view into it.”

Fleet managers can schedule sessions manually, trigger them automatically based on risk profile or use them as structured onboarding. “What we see is that many customers get them at 30, 60 and 90 days after hire, as an example,” Land said.

The output isn’t a raw clip. Managers see a summarized view with video access, and the AI assigns training when warranted. Samsara conducts AI-powered ride-alongs that assess every driver against their proprietary 22-factor standard, the Samsara Safety Assessment Standard.

Customer feedback and testing played an important role.

UNFI was among the early beta testers of Ride Along. Tehzin Chadwick, UNFI’s chief safety officer, noted her team worked with Samsara through the beta alongside UNFI’s operations staff.

That sequencing of test, flag, verify, deploy was deliberate. For Chadwick, driver confidence couldn’t be a soft metric. It was a key part of making sure the change management was in place for the drivers to work with the new technology.

“We aren’t adding technology to make your lives more difficult or more complex. It’s completely the opposite,” she said. “It’s there to continue to keep you safe but also to take away any of that background noise.”

On gamification, Chadwick was direct about the limits of a one-size-fits-all approach. “We specifically try to use that for training. That works with a certain segment of our drivers who really enjoy that. And then there are some other ones that don’t enjoy that, and that’s not how they want to learn,” she said. UNFI now offers drivers a menu of training formats and lets them indicate their preference.

Sanjit Biswas, Samsara’s CEO and co-founder, said that kind of driver buy-in shows up in the data. On leaderboards, competition becomes self-reinforcing.

“A lot of drivers take their safety score super seriously because they’re competing,” Biswas said. “I sometimes do ride-alongs with drivers, and they’re like, ‘Why’d I get a 98?’ And that’s the kind of thing we love to see because it’s a positive feedback loop, and the majority of drivers do a great job every day.”

The old framing, he said, got the incentive structure wrong. “These historical systems had shown that they would call you out on events when you did something bad,” Biswas said.

Voice agents to make driving a less lonely job

The Ride Along feature runs on recorded video. A separate product announced at the conference operates in real time. It is a conversational AI companion for the cab.

The driver briefing agent, currently in beta, speaks to drivers before and during shifts. It can cover safety scores, weather conditions, route notes and policy questions. Fleet managers configure it in plain English through Agent Studio, Samsara’s no-code agent builder.

Land described where the product is headed. “Driving is a really lonely job. You’re out on the roads, driving for many hours, and this can become a companion for drivers,” he said. “It might start sharing some other things that are interesting to them. It might even be really empowering for the driver where they can control this and customize it.”

The regulatory picture is still taking shape. Land noted that the companion agent opens every session with a disclosure: “This is AI on a recorded line.” But the exact requirements vary by state, and Samsara is helping customers work through the specifics.

Biswas acknowledged the product is new ground. “This is a whole new frontier. This has never been possible before for an AI to just be chatting with the driver,” he said. “We want to make sure the drivers feel good about this and that it’s useful.”

For Chadwick, the technology is only as effective as the trust it earns. UNFI’s business, she said, runs on something AI can support but not replace.

“Our business has really been based on trust, relationships, and human instinct,” she said. The technology is there to complement those three pillars.