EP998: How AI is redefining the future of embedded insurance

Ross Sinclair, Founder & CEO, EIP

Global InsurTech funding remained strong in 2025, but what will it take to make sure the hype does not outpace deliverables? Ross Sinclair, Founder and CEO of EIP believes the pressure is now on for many firms to deliver on their ‘pitch deck’ promises, including AI use cases, and justify them to investors with strong revenue streams.

Transcript

SPEAKER_00 0:00

I'm Robin Emler of IBS Intelligence. You're listening to the IBS Live Views podcast. I'm joined today by Ross Sinclair, founder and chief executive officer of embedded insure tech firm EIP. And our topic is how AI is redefining the future of embedded insurance. So let's start at that point. How is AI redefining embedded insurance?

SPEAKER_01 0:30

It's interesting times, Robin, that's for sure. There's an awful lot of snake oil out there, snake oil being sold, should I say. And what I mean by that is there are a lot of companies in our space, the insure tech space, making some fairly outlandish claims about what can and can't be done with AI, particularly in the embedded space. I think, I mean, I tend to get quite animated about this and say some of the cavalier um approaches that companies have to this. But generally speaking, if they're looking at AI application in the industry, you really need to look at how that is being tied back to an objective measurable outcome of some kind. So what I mean by that is normally if you're putting AI into a process somewhere, you should be seeing a reduction in loss ratios, you should be seeing your claim cycle shortened, you should be seeing customer retention levels increase, etc. That doesn't always happen, and you don't always have that evidence available when claims are made in the industry. Um a good example of that is, and I've seen it myself, where a particular business would say that their AI reduces fraud, particularly within things like mobile device, mobile phone insurances with where we work extensively. That AI is good in terms of pattern recognition, and within those types of products, you would look to identify organized fraud, and that's where it does uh serve some purpose because you can see a pattern uh appearing quite quickly, and AI jumps on it. But generally speaking, for individual policies, AI is useless at identifying fraud because you can't you can't see patterns in individual claims, it's just not possible. So, so so yeah, I mean it's useful in certain areas for sure. I would say the AI application has to be properly embedded within a particular product, within a business. And by that I mean don't layer AI onto legacy uh furniture underneath legacy systems. And then probably just recognizing, and we're dealing with this at the moment, recognizing some of this the constraints within the industry as well. And what I mean by that is obviously regulatory constraints, and uh a lot of companies are just ignoring the fact that you can't have AI making claims decisions, um, it data constraints as well, of course. Um it's only as good as the the data that you feed it, and also just customer trust as well, the constraints of customers getting used to AI and and and trusting it.

SPEAKER_00 3:07

Bottom line, AI is not a panacea, but it is useful. How do we stop from what you've said, this is an issue, how do we stop the hype getting carried away uh and leaving the deliverables in the distant rearview mirror?

SPEAKER_01 3:25

It's not easy. Um, as I said, the the the the key the key for me for customers, the advice I would give is back to that point of earlier, which is if claims are being made by a particular provider, tie those claims back to evidenced outcomes. So, yes, you've applied AI, show us what it's done for you, show us the the graph that improves a particular metric or or reduces a particular metric, let's say. Uh that that that would be the key. And and people don't do that often enough at the moment. I think you know, if we look at I I I I saw a wonderful, I saw a wonderful quote just just recently, just a couple of weeks ago, that um talking about embedded insurance, and it said that the the embedded insurance, the best embedded insurance melts into daily life. And I thought that this this was uh this was a wonderful uh quotation because that's what we're trying to achieve. The best forms of embedded cover don't make themselves known until you need to use them. You know, that there's a bit like anti-virus software on your computer. It's uh it doesn't it shouldn't pester you constantly, but just work. Um and AI's helping that a little bit as well, because traditionally, I can give you an example. Going going back in the day, I had a conversation with an elderly relative a while ago, and I've written a blog about this as well, where she talked about renting their television when they were young. And and I said, Don't be that we did as well. My parents rented their TV. And the reason people did that back in the day was not because they didn't have the money for a television uh or the technology was changing dramatically, but it was because they broke down regularly. So you within the television rental to a company called Radio Rental, strangely, you had automatically a repair insurance built in. So if your TV went on the blink in the morning, the engineer would literally be out that afternoon. Now that that that was that was a great form of embedded insurance. It would be repaired or it would be replaced on the spot. Nobody knew that that insurance was in there. It just sat there and waited until it needed to be used. It was the perfect form of embedded insurance. Nowadays, what AI is doing um when it's implemented properly is it's giving relevance at the right moment. So essentially you're matching, it's using data, of course, to match cover to context using pattern recognition usually.

SPEAKER_00 5:51

Okay. Looking at this from the insure tech firm's point of view, not the insurance company's point of view. What do insure techs need to do this year? First of all, to justify all the money they've raised in the last couple of years and the valuations they've raised it at. And I see you grinning.

SPEAKER_01 6:16

Yeah, we're we're we're we're doing it ourselves, we're spending money on on the application of AI. I I saw an interesting statistic from last year from uh MIT um where they they said that uh 95% of companies who had implemented an AI strategy or AI somewhere in the business uh had failed, 95% had failed to see any return on the investment, which is which is a little bit scary. In our industry, it comes back to a point I made earlier. I think that there has to be an understanding that AI can only do so much. And and the particular area there that I mentioned was regulation. Um there's an immutable law within insurance that if it's regulated, then a human signs it. And a lot of the insure techs forget this. And we have where we're doing this, we're bringing this into the business, and we we've sort of vexed long and hard about how to do this properly. But we have a rules engine, and on our platform, we have a rules engine at the back, which is configured by the insurer and then automates the decisioning. So that's that's deterministic. Um, but what we're doing is layering on AI on the front end in the form of an agent, a voice-led agent, which will then act as the servant to that deterministic decision engine. So you have the human sign-off at the back end, you have the human um oversight if you like, but you have the AI doing the heavy lifting on the front end in terms of collecting all the information, answering customer queries, etc. And that's specific to our business. But I think a lot of the insure techs out there, again, as I mentioned before, they need to be aware of the constraints of this industry. It is heavily regulated. Customers have a trust issue with insurance generally, and that needs to be overcome when you start layering on AI as well. So uh and and data. Um, generally speaking, companies struggle with with getting correct data into these models. So that would that those were the areas that I would I would say we need to pay attention to over the coming year.

SPEAKER_00 8:19

Can we therefore describe 2026 as the year of insure tech realization?

SPEAKER_01 8:26

Realization on AI or realization as as as a credible industry? What what what realization in terms of what?

SPEAKER_00 8:35

Well, let me let me ask you to to justify both those points.

SPEAKER_01 8:41

I think this year, particularly this year and next year, 26-27, we're going to see uh a massive amount of consolidation in the insure tech space. At the moment, there are something like four and a half, five thousand insure techs out there, um, which is too many. Um, a lot of them treading on each other's toes. So I think it's it's logical. The ones that have overpromised, and you mentioned investment, taking heavy investment already. I think a lot will struggle to show the returns on that investment. An awful lot will struggle. A lot of them have offered up hyper growth plans, which will never materialize. So I think you're going to see um players disappearing and players consolidating within the industry. I think credibility has to creep into this business as well. Again, back to what I've mentioned already. A lot of the promises being made are just ridiculous. We'd have to find some truth in the in the delivery of services to the insurers. The insurers are getting smarter at buying these services now, unfortunately. Um so I think there'll be a thinning of the a thinning of the herd um over this year and next year, and I think that the promises will become more demonstrable, more evidenced, the technology that's being offered up, some of which is very, very clever. The technology that's that's being offered up will become more nuanced and and uh just a bit smarter as we move forward.

SPEAKER_00 10:06

Okay, a dose of reality there from Ross Sinclair, founder and chief executive officer of InsureTech EIP. Thank you very much, Ross.

SPEAKER_01 10:16

Thanks, Ron.