The Hartford's risk services lead: put computer vision on the cameras a client already owns
Insurance Business's Chris Davis reports on how the carrier uses telematics, sensors, wearables and camera AI to prevent losses before they become claims, and why its risk engineers see productivity gains alongside safety.
Two months before Matt Scott's RISKWORLD interview, a colleague of his at The Hartford laid out the operational side of the same story. Chris Davis's April 22, 2026 piece for Insurance Business profiles Dan Campany, who leads risk services at the carrier, and the argument he makes is that the insurance policy should be the backstop for the unpreventable while everything preventable gets handled with data.
Campany's framing is that the risk-transfer policy exists "for things that are unpredictable or unpreventable." A growing share of commercial losses are neither, in his view.
The four data sources
Davis walks through them. In commercial auto, telematics and in-cab cameras detect drowsiness and distraction. In property, water sensors catch leaks early; Campany's point is that whether a leak is found in a minute or a couple of days "matters a lot" to the size of the loss. In workers' comp, wearables track individual workers' motion and surface patterns a supervisor would never spot by eye, such as a spike in unsafe lifting at a particular hour on a particular day. And on the worksite, computer vision software runs on existing cameras so that AI can watch the video around the clock for unsafe conditions.
That last item is the one this site cares about. The cameras are already there, the software gets added, and the risk engineer's periodic visit turns into what Campany describes as the ability to watch 24 hours a day, every day of the year.
What AI adds to the data
Campany distinguishes the two. Sensors and cameras give granular, objective information. AI finds patterns in that information and then points the human beings at the outliers, which is the same division of labour the video agent vendors describe, arriving from the insurer's side.
The commercial argument he makes to clients is that safety work tends to pay twice. Almost every time the carrier does an ergonomic intervention or reviews telematics data, he says, the client also sees gains in productivity, efficiency and morale. The claim is anecdotal and Davis does not test it.
What the article does not contain is a rate. Nowhere does Campany say a client with camera AI pays less. The Hartford's position, in both this piece and the later Scott interview, is that prevention is a service it provides and that pricing follows the loss experience. My reading, for a plant manager deciding whether to fund a video agent, is that the carrier's risk engineers will help set it up, and that the premium effect, if there is one, arrives later and through the claims record.
Read the original at Insurance Business.