Takeaways:
- Using AI as a research tool is a stronger application
- Predictive modeling is a key application of AI
- Insurtech startups should understand both insurance and AI
Altamont Capital Partners is a private equity firm with investments in the insurance industry including McLarens, a claims management company, and Access Insurance. Joe Zuk, operating partner at the firm, focuses on investments in property and casualty, and accident and health insurance, backing startup and mid-size companies in those businesses. He looks for companies that take a digital approach to insurance operations, and more recently, those applying AI to the business. Digital Insurance spoke with Zuk about insurers' best bets for using AI.
This article is from a longer interview and edited for clarity.
What applications of AI for underwriting are making the most progress?
The strongest that we're seeing, or that I'm seeing across portfolio companies and just in general, tends to be really on AI as a research tool. Plus copilot, specifically to help organize and dive into just vast amounts of data. AI has been most effective around ingestion and sorting. There are some good use cases, whether it's visual or in documents where they're coming across interesting pieces of information and flagging it to a human's attention. That's where we've seen some effective use of AI.
What kind of data is AI being applied to, or should it be applied to, like risk data?
Exposure and experience data. Exposure data seems to be still the hardest, the long pole in the tent. Experience data is a bit better, but still, there's work to be done.
What are visual applications of AI in the insurance space and why are they valuable?
We've worked with a vendor on this, taking imagery and utilizing that to determine minute changes or nuances of detail that maybe a human won't pick up, in terms of changes in exposure. Or, post-event analyzing and saying, 'Is that literally possible or not right in terms of a car accident or something to that effect?' That's where we've seen a little bit of AI.
The other use of AI would be predictive modeling. Taking historical events like weather or natural perils, and projecting the probability that a winter storm event will happen. What is the probability in a given time frame, and what are the probable damage assessments?
What concerns you about the insurtech market, or what insurtechs are offering?
Lack of knowledge in the insurance space. Arrogance, ignorance. [Insurtechs] come in and think insurance is easy. They don't understand the regulatory nuances or product program. I'm not quite convinced that consumers want brokers to completely go away.
What do you want to see from startups pitching Altamont Capital?
We're looking for two things. One, true domain expertise in insurance or the element that they're trying to solve for in the value chain. Two, deep understanding of AI and technology and how that applies. Three, a team that's assembled from both professional experience or credibility, but with diversity. So it's not just purely insurance people. It's not purely just tech people. There's some well roundedness in the team, and some seasoning.
What does the outlook for investment in insurtech look like for this year?
The outlook here last year was down. There's still money to be found. It's a little bit harder. It depends on macros like where interest rates go, and liquidity. Also, probably larger factors in the capital markets than necessarily folks are paying attention to.
There's a lot of interest in AI generally, and a lot of funds flowing in. Yet at the same time, there's obviously macro questions as to whether it's real. AI is real. But are the returns really there for investors?
I think AI technology is here to stay. Like everything, it will ebb and flow.
