Insurance CVCs are eyeing up the potential of AI even if regulation may make adoption slow. Climate risk mitigation and quantum are also on the menu. Blockchain — not so much.
The recent bear market has shifted what corporate investors in the insurance sector are looking for in startups.
“Larger enterprises and larger traditional insurers have been feeling the pinch from inflation and everything else over the last 18 months. As a result of that, R&D, a lot of proof of concept and innovation work has been axed to cut costs. That all filters down into the insurtech ecosystem and on to those startups,” says Francis Lowry, senior associate at AllState Strategic Ventures.
Insurance startups themselves have changed much of their approach — last week we wrote about how they have moved from trying to disrupt incumbents to seeking a new partnership models with them. Meanwhile, corporate venture capital investors have honed what they are looking for.
Gen AI could be huge in insurance — but nobody is sure how yet
“The one that hasn’t been cut within enterprises is the AI budget,” Lowry adds. “In a lot of cases, it’s been increased. So, there are a lot of companies with good artificial intelligence components to their solution that have access to good proprietary data. Those companies are weathering the storm and are actually increasing their traction.”
Insurance investors see a lot of potential for incorporating generative AI into their industry, but don’t necessarily agree how. AI could, for example, make claims settlement and adjudication faster, but the biggest potential gain is applying it to the customer side – ‘the golden area we all want to get to’ as one insurance CVC described it.
However, as long as generative AI is prone to “hallucinating”, or generating nonsensical or inaccurate outputs, the risks outweigh the financial benefits. Regulatory oversight means the technology may well be restricted to being an assistance tool for the foreseeable future.
“The thing about insurance, like most fintech, is that you can’t move fast and break things”
Eugenio Gonzalez, Plug and Play
“The thing about insurance, like most fintech, is that you can’t move fast and break things, otherwise the FCC will come in and shut you down,” says Eugenio Gonzalez, who heads venture innovation platform Plug and Play’s insurtech investments. “A lot of companies learn that the hard way. It is a highly regulated area, and with good reason.
“The main issue with chatbots is you need them to be incredibly accurate when they’re facing your customers. A mistake there can be very expensive,” he says. However, Gonzalez believes that AI could be deployed more rapidly with staff.
“I do think a lot of companies are testing out solutions, most likely internally, for knowledge management for instance. You can have an agent or chatbot that can talk to and basically train your team or help them navigate certain items, where you are basically able to now unlock a tutor that’s available 24/7,” he says.
Plug and Play has invested in an Amazon-backed startup called Archetype, which pulls data from different sensors – a Ring doorbell camera for example – and uses AI to better understand its context. If you’re a claims adjuster for instance, you could apply it to a video feed to get a clearer idea of what’s happened in a car accident.
Still room for small AI startups
The sector can sometimes be seen as a difficult environment for insurance AI startups, because the insurance industry has been using AI to measure risk for years and many of the biggest players already have large teams developing their own internal models.
But there is still room for startup technologies, says Steve Bernadez, who spent five years as a partner at CSAA’s corporate venture arm, Avanta Ventures.
“We do like to keep our own models and algorithms to ourselves, but we still need the infrastructure to run those models, in increasingly modernising stacks,” he says.
“Do we have the ability to do [machine learning operations], model ops, data governance, all of those things [we need] to run modern models? Because there’s a lot of innovation going on in that infrastructure beneath analytics, and that for me is ripe for investment by insurance CVCs.
“The other thing is that there are different types of insurance companies. There are your State Farms and AllStates of the world which are big enough to have hundreds and hundreds of data scientists, and then there are the middle-market insurance companies that don’t. So, they will need people to help them build those models, and they need to rely on service providers or data infrastructure providers to help them do that.”
Climate change shifts insurance investors focus
Climate change is the other issue on every insurance investor’s mind as it is already wreaking havoc on insurance company balance sheets. Worldwide insured losses for earthquakes and climate change-driven storms topped $220bn in the past two years alone, according to reinsurance firm Munich Re.
“You’re seeing a lot of VC dollars going into sustainability, resilience and everything that has to do with that,” says Gonzalez. “For insurance companies, the increasing severity and frequency of these natural weather events are having a huge impact on the books. Property and casualty insurers are suffering from wildfires, incredible storms, hail in the Midwest, tornadoes and so on.”
In the US, some firms have stopped offering home insurance altogether in volatile states like California or Florida. But packing up and leaving isn’t an option if, for instance, you’re a Japanese company and your home market is earthquake-ridden Japan. And so more money is flowing into startups that can both predict and mitigate climate risk.
That is subtly expanding insurers’ business model from pure risk protection to proactive risk management, says James Orchard, CEO of QBE Ventures. Insurance carriers are now also trying to help customers understand the biggest risks to their business, and working with them on strategies to combat that risk.
“That hopefully ends up reducing claims and helps our customers become more resilient,” he says. “It’s one of things QBE is looking at: how do we help them understand the impact of climate change on their supply chain risk for example? How will that affect their business and how can they start mitigating some of those risks?
“We’re also seeing products that have been around for some time now starting to get more headway,” Orchard adds. “Products like parametric insurance; some insurers have stepped away from some risks in some markets, and we’re now seeing new technology and product structures enter those markets with new value propositions that may be more fit for purpose for areas that are super prone to weather events.”
Insurers are exploring uses for quantum computing
Insurance investors are also looking further ahead into areas of deeptech.
“We’re making some quantum computing investments now, which would surprise a lot of people,” says Michael Remmes, who leads insurance group State Farm’s venture office.
“There are a few use cases we’re thinking about. If we can incorporate weather data into insurance modelling and maybe run our whole book of business through something like a certain hurricane scenario – these could be real or hypothetical events – quantum might be able to help there.
“We invested in a company in Singapore that focuses on error correction, and that is going to be very important for the future. Quantum computers give off a lot of errors now – you have to tame those qubits. If they give off a lot of wrong answers, nobody’s going to trust them, and you can’t really sell a quantum computer it it’s not trustworthy.”
Blockchain out of favour
One technology area that is off the table now for many insurers is blockchain. Five years ago, says AllState’s Lowry, it looked as if blockchain and smart contract technology was a possible answer to simplifying claims. In the event of a car crash, each party’s insurance company needs to agree on a payout, and that takes time and money. A smart contract system could potentially do that automatically.
“The problem is that your insurer, my insurer and probably 10 others need to buy into the same platform,” he says, and no one could all agree to use a particular platform. “Because you can’t be on five different blockchains, you have to all be on the same one in order for that to happen.
“Ultimately, if you can’t get that consensus, a centralised database system will still do the same thing for you. The world isn’t going to end tomorrow if you don’t use blockchain, we just revert to using the centralised data system we use now where transactions speak between the insurers.
“We didn’t actually do anything with blockchain,” Lowry adds. “It was one of those hype technologies at the time. Generative AI is different, it’s gone past the hype and we can see the real value in the technology.”
