Building the Future of Risk Monitoring with Dominick Zangaro

Dominic, as we, as we take a step back, though, I mean, uh, y- you've been in the industry for a few years now.

W- how would you define the current state of data and AI as it pertains to insurance?

I think the amount of data is massive.

I think the actionable data is difficult to utilize right now, if I summed it up into one sentence.

I can't speak for how other insurance companies do things per se.

I, I, I feel Progressive probably is one of those insurance companies that has a massive data lake and can probably act on the majority of that mass.

Um, and I'm sure that there are, you know, small mutuals that are 10 to 15 million in gross written premium every year that probably have a massive data lake that they can't actually act on it, right? Um, and I think that disparity creates opportunity, one, but, um, I do think that there is an element of, um, the insurance industry does have a massive hurdle to get over around how do we make the data that we have available, and how do we create an artificial intelligence system or leverage a large language model or provide agents with that data and actually expect them to be able to produce the results that we need? Um, that's probably the most difficult jump that the industry is going to have to make, is the efficacy of any individual model.

The hallucinations that artificial intelligence can provide are drastic, and so, um, trying to just strictly rely on it from, like, a data reporting perspective or decision-making perspective can be difficult and, and that's kind of why I'm throwing my eggs into the basket of automation and being able to allow agents to be able to run different procedures rather than saying, "Hey, here's an agent.

Let them make a bunch of decisions and tell you what your finances are gonna be like in, in 10 years," right? I don't feel like actuaries are going anywhere in the n- near to even long term.

Um, I, I think that is so sophisticated that agents and artificial intelligence are probably, um, not the right fit, but I do think there are elements that we can improve on.

Um, one, I would say the distribution model is probably already sufficient.

Most organizations understand that they can go online to a website and get a quote.

Um, most individuals definitely understand that, and when you look at somebody like Lemonade Insurance, they already-- You know, that, that distribution model has existed, uh, for 10 years.

I'm not sure artificial intelligence or data is going to continue improving that.

But I do think on the claims side, there is definite opportunity for, uh, more speed when an actual customer wants to use the product, and I think that's where, uh, more organizations will look to provide investment, um, and if they're not, I think that's where they should look to, to put investment.