The Hartford's IoT executive talks tech use for prediction, prevention

We're trying to use connected devices and the data from connected devices to mitigate insurance claims. In other words, things that are preventable and predictable, let's try to address them, and prevent and predict them, so that they don't happen in the first place. We're using that data to improve the granularity of exposure information to do a better job of underwriting and put a fairer, more equitable price on the risk by understanding the nature of that risk better. The lab creates novel new products and customer value propositions, leveraging devices and data such as usage-based insurance.

To prioritize which types of technology and which customers to work with, we look at our loss data and understand what types of losses our customers are experiencing that we believe could be preventable or could be predicted. As a result of that analysis, we've prioritized sensors like water damage prevention technology, both at a personal level in the home and at a commercial level on construction sites, large office buildings, schools and universities. Each of those requires nuance in the technology and how we work with those customers to set up an effective water damage prevention technology system.

In addition to water damage prevention, we are looking at worker safety, which includes wearables for ergonomics. It also includes computer vision, where artificial intelligence can effectively watch video streams and identify risky things like forklifts that are doing donuts in the warehouse or workers who are jumping over assembly lines. We can use those insights to work with those customers to help make their operations safer for their employees.

We are also looking at commercial fleet telematics. Most think of telematics as a personal lines thing today. But we believe there's a ton of opportunity in commercial fleets. That technology includes not just the data that you can get off a mobile phone app, but also includes in-cab cameras that face both at the driver and out towards the road to understand the real-time conditions. Is the driver distracted or drowsy? What kind of traffic conditions are they facing or weather conditions are they facing? We can use that data in a lot of different ways both in helping them be safer, but also if a claim does happen, understand what happened, why and who's at fault.

The way we are set up, the IoT lab is designed to be an incubator of ideas. We source the ideas from our own research, from our loss data, from our underwriters, from our risk engineers. We take those ideas and frame them into a vision. We extract the key hypotheses for the idea to be successful at a meaningful scale. Then we experiment on a smaller scale with five, 10, 50 or 100 customers to really test the idea in the marketplace.

If that is successful and we can figure out the process, the economics, the technology and the customer experience, we take those ideas back to the business units as '90% baked,' something compelling to scale and operationalize within the business. We work with them to transition that capability into the core business. Sometimes that's as simple as risk mitigation and using our risk engineers to go out and install devices. In other cases, that's, 'Hey, we're getting telematics data and we think that can be used for pricing segmentation.' That gets more into the risk selection, the policy structure and how we use it in our rating models.