Digital Insurance spoke with Sarah Kim, a partner at Centana Growth Partners, a growth equity fund focused on financial services, about the current market and the
The conversation has been lightly edited for clarity.
Can you share what you're seeing in the market?
One, it's very dynamic. We've been seeing a lot of people come back to the market to fundraise. They're not kicking the can down the road as much, hoping that the market will get better. But again, public markets are changing again. And so I just say it's a little unpredictable. There's a lot of excitement around new technology, people trying them. And yet, there's also a question about how it will be implemented. So I'd say that overall creates excitement, but it also creates some let's see how it really plays out.
I'd say in M&A, we have actually seen a few things here and there. While last year was a real low, we're also seeing some interesting activity where some smaller players might be combining. MGAs might be attractive acquisitions. So, I wouldn't say it's dead. I'd say it's an interesting market. The big players are buying software. They're trying them out. They're testing things out. Existing startups are leveraging AI to speed up development and deployment and become more efficient. And you see, the big carriers and the brokers want to leverage the software to also become more efficient. So I think the market is still functioning. But again, I wouldn't want to predict anything, given all the changes that happen about every day.
What types of startups are you most excited about now?
I think, across distribution, underwriting, carrier capacity and claims, we're seeing lots of interesting developments.
One, definitely on the underwriting side, leveraging a lot more data. How can you leverage unstructured and structured data to have more precision in your risk analytics and underwriting? Everything from being able to take a first notice of loss and automate so many developments on that front. And then, because of the volatility in the insurance market, on the carrier capacity side, we're actually seeing a lot of dynamic markets. So, who should workers go to? Who should agents go to? How can they access and find capacity? How do you help independent agents serve their customers when you have carriers not working really well in California, for example. Some new carriers are coming in, taking huge losses right after the LA fires, but others are wanting to step in.
How do you see technology being integrated into these areas?
Leveraging everything from vision recognition and LIDAR to help prevent accidents and workers comp issues in the workplace.
As you think about unstructured data historically, a lot of our catastrophe models, a lot of our loss ratio models, are historical data and trying to trend it. Now, we have access to so much more unstructured data that we can pull in and have a more forward looking, predictive, specific and precise risk model, versus it being higher level, more generalized, like they had to historically. And so we're really excited to see things like that on that front, unstructured data, visual data, everything coming together for more micro underwriting versus the broad swath of portfolio level underwriting.
The great thing about carriers and underwriters is they want more data, not less, right? And so it's actually a little bit of a smorgasbord of more data. You have to prioritize and figure out how to make it applicable, but it can bring insights faster, less time cleaning data, processing data, and more time actually analyzing it, assessing risk and making the rice risk selections.
Do you anticipate AI to reshape how people do their job within insurance companies?
One thing about the insurance industry is it hasn't always attracted the next generation of talent, and so we actually need to be able to scale our operations, leveraging technology, in particular, things like customer service and claims management. We've seen with some of our portfolio companies that are already leveraging AI to reduce inbound calls and to really provide better customer support. So those tools on the front-end are helping.
And then on claims management, we're seeing some really great innovation with startups on claims management, and that's really important, because we don't have a lot of claims adjusters out there. We don't want to send people out in the field. So how do you leverage technology and AI to streamline those processes? And we're seeing everything from not just processing the documentation, but actually AI bringing new insights, insights on reserving, on claim settlement paths, on how to recognize issues earlier with a claim.
And so I think we're seeing it being a complement to the human side. So instead of having to add bodies behind things we actually create, they're leveraging AI agents to complement and to support the humans. Our claim adjusters and our underwriters can do the higher level thinking and risk selection, risk management, but leverage AI for insights, for cleaning all that unstructured data. Every day, it makes you much more efficient. Turnaround times are increasing, whether it's access to carriers more quickly, whether it's access to your check if you have a claim, faster. All these processes, I think we're seeing leaps and bounds of efficiency gains.
What challenges arise in upskilling workers to use new technologies?
I think because a lot of it's based on natural language models, now, it's actually easier in a lot of ways because you can have AI agents that interact with more of a colloquial learning system. You don't have to go back and train people on specific software use cases, and every time you upgrade, retrain. That training can happen in the background, and the interface is actually talking to another colleague. I think that's where the agentic AI actually, really comes in and makes it more useful, more practical, regardless of which portion of the workforce you're working with.
If you can type into an agent or the question from customer service comes straight into the agent first, it creates the first line recommendation or response. Then your customer service agent can escalate when there are questions that need a more complex understanding. It's right there, they don't have to go into four different systems. They don't have to go ask five different colleagues. That model is pulling all that knowledge together, and it's at the tip of your fingers. And so I actually think it's easier in a lot of ways.
I do think people have to get used to how to use AI, how to ask the right questions and the prompts or training models on your industry specific needs are very important. Human interaction is actually easier. And you know, one of our portfolio companies actually took some of the standard off the shelf large language models (LLM) but then took some models and trained them on customer service calls, materials. The response rate and efficacy was leaps and bounds above the base models. So trained on the company's materials, the company's expertise, and then you can get that out to the front line of customer service and make them each more effective. And so we're already seeing the huge step up in productivity.
How are companies using Agentic AI?
We're seeing a lot of pilots and use cases at a lot of point solutions. First of all, I think what we're still waiting to see is if there's going to be a full stack system replacement. And I think insurance companies are cautious. They don't jump in all at once.
Say, a company has a claims management system, they might have built. They might layer on an agent solution, but not replace the full workflow. We're also seeing companies that start that way and then go, this is crazy, I'm just going to replace it and use your entire claims management system, because my existing one just can't adapt that fast. So, I need one that actually can automate the entire workflow, create new workflows that are actually more efficient, eliminate old workflows that don't need to happen, can happen in the background and be automated. I think we're seeing an evolution. I think it's going to be dependent on the carrier and broker. How comfortable they are. But I do think the speed at which we're seeing change is very exciting. And I honestly think the legacy players and the largest players in the world need to be at the forefront of that. They do have budgets. They're spending money. We'd love that to continue. We hope no hiccups in the economy kind of slow that down. But I think people realize that the game is changing very fast, and you need to be out there testing things, find your right solution, and eventually upgrade all of those systems over time.
Because we are using so many new models, leveraging LLM, leveraging quantitative AI models, I think you are going to continue to see pressure on compliance on the regulatory front to make sure that there's transparency, there's auditability, and that you're leveraging that predictive power, but that the consumer, that end customers still being treated well. And so we're going to see regulation constantly, state by state, also play a role here, and I think we're still feeling our way through on how that's going to work with AI.
Do you anticipate M&A activity to continue?
I don't know if there's going to be a lot of big, mega deals. I think there's going to be a lot of small deals. I think there's smaller startups that want to consolidate and partner together to get more scale or accelerate product roadmaps. So, I do see a lot of that. I actually think, for the platforms that have survived, that are doing well, and then there's platforms that have some interesting technology, but maybe have not hit escape velocity on size. I think combining some of those are going to make some more formidable competitors, and we're going to see some consolidation. I think that's probably lower in the market, on the MGA side.
I think, like always, we're going to see brokers and MGAs merge, start new ones. I don't think that cycle is ever going to end. And I think as you have new MGA players or players that are leveraging technology for underwriting and distribution, they'll be great acquisitions for brokers or carriers alike.
