Paul Carroll
What are the main challenges in underwriting today, and how might emerging technologies address these issues?
Balázs Kaman
One of the biggest challenges in underwriting today is simply getting the right data into the system. Many of our MGA customers underwrite highly specialized risks such as crypto exchanges, mining rigs submerged in the ocean, or electric vehicle chargers. The prerequisite for accurate underwriting is having high-quality data available for rating, but collecting and structuring that data is still painful and time-consuming.
Over the last decade, the industry tried to solve this by pushing data entry downstream and asking agents or insureds to fill out information directly in portals. With AI, we now have a better path forward. Instead of changing long-standing submission behaviors like sending emails, AI can extract and structure that data automatically and trigger the necessary workflow steps. That allows underwriters to spend less time rekeying information and more time focusing on evaluating risk.
Paul Carroll
I’d bet that the speed enabled by AI creates benefits beyond efficiency. In the consulting world, the concept of "time to value" has really taken hold over the past 15 years. In other words, don’t just tell me you’ll double my investment; tell me whether you’ll do that in two years or 10. What value does this acceleration bring to the underwriting field?
Balázs Kaman
Absolutely. Tools that integrate via APIs [application programming interfaces], enriching the data that you have available, using AI to do some of the underwriting and also scoring the incoming requests let you focus on what really matters: These tools enable underwriters to cut processing times dramatically.
What used to take maybe days can now be handled immediately or can at least surface a preliminary price or rate, so you can then come back with a more polished rate after all the underwriting was taken into consideration.
Paul Carroll
I imagine faster quote turnarounds provide a competitive advantage in the highly competitive MGA market?
Balázs Kaman
Speed is the name of the game. We see that in all the customer types we serve. Wholesale brokers who don't necessarily do the underwriting themselves but focus on finding the markets that need a specific application—speed is very, very important for them. Also for MGAs who rate their own risk and do their own underwriting and who might have binding authority.
In the past, they used platforms where they needed to log in and rekey all the information. Modern systems like BindHQ allow integration with their APIs directly and massively reduce the time it takes to get quotes back.
Paul Carroll
How has insurance evolved in the six or seven years since BindHQ was founded?
Balázs Kaman
I can say that many of the big frustrations—issues like duplicate data entry, disconnected systems, and long response times to customers—can all be addressed with today's technology.
Previously, only a handful of forward-thinking carrier markets had APIs, and even those weren't very modern or helpful. They typically only allowed for quoting, not binding or endorsing policies.
Our industry is slow to adapt. Still not every carrier has those capabilities. But we are getting there.
Seven years ago, I saw a lot of handwritten, scanned paper documents being uploaded into agency management systems. Then someone, mostly offshore, would rekey that information. Today, modern OCR or AI-assisted tools can read and process that information, which saves time, reduces cost, and creates a much better user experience.
Seven years ago, people simply sent ACORD forms in emails. This practice is still fairly common because people resist change. They wonder, “What's in it for me?” You really need to provide incentives to agents to start using new technology or platforms, or they’ll just email the expiring policy or a filled-in ACORD form from two years ago.
If you tell agents to come to your platform and rekey everything, that won't work. But with the new tools using OCR and GenAI to extract that information, you can save them tremendous time. As an MGA, if you can save time for retail agents and quickly provide accurate quotes, they're more likely to send business to you. I find it amazing how long forms have remained relevant despite technological advances.
Paul Carroll
I hear all the time about problems with data standardization in the insurance industry. How do we address that, given that data is expressed in different formats across different systems?
Balázs Kaman
The question is tricky because insurance is complex. I joke that specialty insurance is anything about anything. It's a written contract about literally anything. So there are either no standards or there are too many standards.
People have tried coming up with standards, but specialty MGA program administrators come up with creative and innovative products, so how they capture data might change. And depending on who’s viewing the report, they might want to see things differently. So we provide access to the data and allow you to really build your own report.
There, again, generative AI can be hugely helpful because the tools can really democratize the data engineers' work. You can, in plain English, explain what reports you want to see. And then if the data is available, you can more easily build those reports.
But GenAI is unfortunately not a silver bullet. You cannot just put ChatGPT on top of a database and expect it will solve all the problems, because insurance is very complex. Depending on how you ask the questions, you might get different answers. Like, are you thinking about the term premium, the billed premium, the annual term premium, the pro rata premium? Even just "premium" has so many meanings that you need to be very careful when you are creating a report.
Paul Carroll
Where do you see underwriting heading over the next two or three years?
Balázs Kaman
That is a great question. I've read a quote that people usually overestimate change in the short term and underestimate change in the long term. I think the GenAI hype has maybe settled down a bit. Everyone started using it, and they burned themselves once or twice. Now, some people are saying it's not the revolutionary thing we thought it would be.
But even if we just implement everywhere in all the tools, in all the workflows, the technology that is available today, it would already mean a huge change for the entire industry. Automating the busy work will be huge.
Also, I think concepts that are not even considered today will become more prevalent. Using AI agents and building custom agents to do underwriting and enrich data will be huge. Accessibility and the interconnected nature of our industry will be better.
I think the trick with GenAI is that you don't need to change human behavior. You can just put smart tools on top to get huge results.
I still think insurance is a relationship business. There will still be a huge role for the relationship part and the human element. Technology will not solve all parts of the problem, like securing capacity.
Paul Carroll
How accurate is AI, and how accurate does it need to be?
Balázs Kaman
97% accuracy is sometimes great, but sometimes it's not good enough. If you need to report on your financials, for instance, 97% accuracy is not sufficient. However, if you want to provide speedy responses and a ballpark estimate is acceptable, then 97% can be good enough.
I think that's where the difficulty lies for many technology providers. Getting from zero to 95% accuracy is pretty easy with these new technologies. But going from 95% to 100%, where you can totally trust the system and take the human out of the process, is difficult.
Paul Carroll
We’ve published numerous articles on “continuous underwriting,” where companies adjust policies in real time when conditions change rather than waiting for renewal periods. How do you view that concept?
Balázs Kaman
The technology would allow that, and really forward-thinking companies are doing it. In business auto, it's very common that you continuously underwrite. Based on the mileage that was driven, you fine tune the policy and dynamically do the rating. I believe this trend will intensify, particularly with the proliferation of IoT, as ubiquitous connectivity becomes the norm.
I also see embedded insurance as a trend, building on that connectivity through APIs. With AI, the integration part can be much simpler.
Paul Carroll
Any final thoughts?
Balázs Kaman
I think this is a very exciting time. In the past, humans were afraid when there was change, thinking they would be out of a job. But the coming years will help underwriters reduce the busy work, the boring stuff—keying in information, sending emails, responding to emails—so they can focus on what requires their expertise and on the art part of underwriting. The boring stuff will be taken care of by technology.
About Balázs Kaman
 | As Head of Product at BindHQ, Balázs leads the company’s product strategy and innovation roadmap, shaping how MGAs, wholesalers, and retailers connect through BindHQ’s modern insurance distribution platform. He champions customer-centric design and scalable architecture, ensuring products deliver measurable value for underwriters, brokers, and partners. With over 12 years of experience spanning product management, software engineering, and business operations across multiple indusries, Balázs combines deep technical expertise with a strong commercial mindset to drive digital transformation across the insurance value chain. |
