In part 2 of our conversation with Nauto’s CEO Stefan Heck, we discuss the measurable changes being driven by AI safety implementation in commercial fleets today. 

This interview has been edited for length and clarity.

Q: What does “in-the-moment coaching” look like in practice, and how do drivers typically respond to it?

Heck: In practice, in-the-moment coaching means delivering real-time, in-cab feedback that prompts drivers to correct risky behavior as it is happening. 

For example, if a driver is following too closely, is looking away from the road, or engaging in distracted behavior, the system provides an immediate audio or visual cue so the driver can adjust instantly. 

This shifts safety from after-the-fact review to live intervention, without continuous recording, because most data is processed in real time rather than stored.

Drivers typically respond positively when the system is accurate and not overly intrusive. At Nauto, we also use a range of different interventions from nudges to voice coaching to alarms, depending on the severity and urgency of the risk and whether the driver continues to engage in risky behavior. 

Because the driver knows they are not being continuously watched, but only alerted in meaningful risk situations, trust builds quickly. Alerts are context-aware and triggered only when risk is meaningful, so drivers often self-correct without the need for manager follow-up. 

Q: What long-term behavioral changes are fleets seeing after implementing AI-based coaching systems?

Heck: Over the long term, fleets see sustained improvements of 80-90% reductions in risky driving behaviors, particularly reductions in distracted driving, tailgating, and other high-risk actions. 

Because AI provides consistent, real-time feedback, drivers develop safer habits that persist even without active alerts - we have measured that this lower risk driving carries over into their personal vehicles. 

This leads to a measurable shift from reactive correction to proactive awareness, where drivers anticipate risk rather than respond to it.

In addition, fleets often report a cultural change around safety. Drivers become more engaged and accountable, while trust in the system increases due to its accuracy and non-intrusive nature. Over time, this results in fewer interventions needed from managers, more consistent safety performance across the fleet, and a lasting reduction in both incident frequency and severity.

Q: With distracted driving responsible for a majority of collision losses, how are insurers responding to fleets that deploy AI safety systems?

Heck: Insurers are just starting to differentiate much more clearly between fleets based on how actively they manage risk. 

We still see some insurers offering generic 5% “telematics discounts” without looking closely at whether these systems can detect/track, let alone prevent distraction, which drives 70% of loss. 

A telematics system that only detects hard braking and speeding will NOT yield material savings of more than 5-10%. 

Fleets deploying AI safety systems and demonstrating sustained reductions in distracted driving and collision frequency are gradually being viewed as higher-quality risks, and that is beginning to translate into tangible outcomes. At renewal, these fleets are seeing more competitive pricing, stronger insurer appetite, and greater flexibility in program design, driven by confidence stemming from evidence of reductions in both frequency and severity.

It also has a material impact on access to capacity, particularly in segments where placement has historically been more challenging. 

Fleets that can demonstrate consistent, data-backed improvements in safety performance are attracting broader insurer participation and more stable long-term support, which, in turn, strengthens their overall negotiating position. 

As this shift continues, leading insurers like Progressive are placing greater weight on demonstrated safety outcomes, and fleets that invest in AI-driven risk reduction are beginning to see those investments reflected directly in their insurance results, both in cost and in the quality of coverage they can secure. But there is a long way to go.

Q: Can AI-driven insights be used defensibly in litigation or claims management?

Heck: Absolutely - AI-driven insights are increasingly being used in both claims management and litigation. The combination of video with behavioral and contextual information provides a clearer, more complete understanding of events leading up to and during an incident, enabling insurers and fleets to assess liability with greater speed and confidence. 

In practice, this improves claims outcomes through faster resolution, practically eliminating fraud, reducing disputes, and strengthening evidence when claims are challenged. 

Commercial drivers, on average, are hit more often than they cause collisions, so they will be exonerated by AI systems more often than incriminated. A good predictive AI system will also warn the driver 3-5 seconds before a collision, giving the driver time to avoid many at-fault collisions. 

The defensibility of these insights depends on how the data is captured, governed, and presented, with a clear focus on auditability, transparency, and compliance with privacy and regulatory frameworks. When those foundations are in place, AI becomes a reliable source of evidence to support decision-making throughout the claims lifecycle. 

Q: What challenges do large, multi-region fleets face when scaling AI safety solutions across thousands of vehicles?

Heck: The main challenge is building driver trust at scale. 

In large, multi-region fleets, adoption can stall if drivers feel they are being constantly monitored or unfairly evaluated. That is why it is critical to design systems that are accurate, non-intrusive, predictive, help drivers prevent collisions rather than just record them, and privacy aware. 

By avoiding continuous recording and focusing only on meaningful risk events, AI can support drivers rather than surveil them. This approach helps build trust quickly, which is essential for successful adoption and long-term impact across diverse fleets. 

For very large fleets of over 10,000 vehicles, installation can also be a challenge, since an installer, an AI system, and a vehicle must be scheduled to be in the same place at the same time. Nauto works with vehicle manufacturers and upfitters to make this much easier for new vehicles, which can simply be delivered with Nauto already preinstalled. 

Q: What advancements in AI-driven safety should commercial fleets expect over the next 3–5 years?

Heck: Over the next 3 to 5 years, what is considered advanced today, such as proactive and predictive safety, will become the industry standard. 

AI-driven systems will move beyond reactive monitoring entirely, with fleets expecting real-time risk prediction based on a combination of driver behavior, vehicle signals, and environmental context.

We will also see greater consolidation into unified platforms that bring safety, operations, and compliance into a single view, enabling faster and more scalable decision-making. 

At the same time, AI will become more personalized, continuously adapting to individual drivers and improving coaching effectiveness over time. The AI will also migrate from an aftermarket add-on to be built into new vehicles. 

Over time, it will not only warn, but also take action - braking, swerving, or avoiding pedestrians actively when the driver has not or cannot respond.