DICEUS CEO Illia Pinchuk on the RiskCellar podcast
DICEUS CEO Illia Pinchuk joined Brandon Schuh and Nick Hartmann on the RiskCellar podcast to discuss how AI is changing insurance software and what this means for brokers, MGAs, carriers, and captives.
The conversation focused on a practical question: how can insurance businesses make their operations faster and less expensive to change?
Illia and the hosts explored several shifts:
- From per-seat pricing to value- and outcome-based models. The discussion looked at how software pricing can reflect the value delivered to an insurance business.
- From complex technology stacks to more connected workflows. Reducing friction across legacy systems remains an important part of improving operations.
- From lengthy changes to configurable platforms. End-to-end platforms can help businesses adapt workflows faster as their needs evolve.
- From AI adoption to controlled enterprise use. Clear controls, traceability, and audit trails are essential when introducing AI into insurance operations.
The episode also covered how insurance organizations can scale without a proportional increase in software costs and operational complexity. Listen to the full episode:
Here’s the transcript.
Brandon: Hey, RiskCellar listeners, this is the interview segment of the episode, and I’m here today with Illia Pinchuk from DICEUS. Illia, thanks for being on RiskCellar.
Illia: Thank you very much for inviting me. Happy to be here.
Brandon: DICEUS is in the software business, and as we go through a lot of different iterations, ideas, and prophecies around the future of software versus AI, I wanted to have Illia here to talk about what they’re doing in the space and how they’re competing. So, Illia, why don’t you dive right into your background? How did you get here? Give us the Reader’s Digest version of your biography.
Illia: Thank you again for having me today. I’m the founder and CEO of DICEUS. I started this business in 2011, and my educational background is in robotics and artificial intelligence. So technology and AI have been part of my interest from the very beginning of my career. Over the years, DICEUS has developed a strong focus on insurance.
As for how I started this business, I can’t quite call it a coincidence, but one of my good friends invited me to build something nice for the insurance space, since at that time I had been working for many years for a few quite large global insurance organizations. He said, let’s build something interesting. We built quite a nice project, and I thought: why don’t I just go ahead and create my own company? So I started my own company as a services company, providing tech services for insurance companies. We also worked a bit with the banking sector at that time.
Then, after some time, I thought: why not go ahead and create our own ready-made products? That’s what we did, and we started providing them to carriers, MGAs, and brokers. We also have a growing focus on the alternative risk market, particularly captives and self-insured organizations. Our work covers policy administration, underwriting, claims, the general ledger, financial operations, reporting, and the workflows around them.
I personally have spent many years looking at how technology can solve real operational problems in insurance. And what I’m particularly interested in today is not just how AI can automate existing work, but whether it can actually make the insurance business faster and less expensive to change.
Brandon: So where do you see the market right now? Obviously, there’s a lot happening. We’ll talk about some of the products that your clients’ clients are probably starting to see come out right now. Meta just announced this Muse product last week, and it’s had tremendous interest, both from the equity markets and from a lot of prophetic-style views, crystal balls into the insurance market. But where do you see your competitive space right now, and where do you play in that spectrum?
Illia: First of all, we have a significantly different business model. And I envision this is the model that will develop rapidly in the market. The idea behind it is not to price per seat. We don’t charge our clients per seat. We charge based on the value and the outcome we provide to our clients. At the end of the day, it doesn’t matter to us how many employees will use our software. Even if, at the end of the day, a single employee uses our different AI agents connected to our ecosystem and orchestrates a large insurance carrier, they will still pay the same. So it doesn’t matter to us how big the organization is. What we value first of all in our business is the value and the impact we bring to the client’s business. That’s the first thing.
The second thing is that we obviously focus on building configurable solutions and software that can be easily configured by clicking a few buttons, instead of asking many teams, many developers, and different employees inside the company to change different subsystems.
And the third thing is that we develop these solutions, or actually not one solution but multiple solutions, as end-to-end platforms. That means you don’t have to go to different providers or internal teams to configure, reconfigure, or develop software. You can do it easily in one solution. That’s our main focus. And of course, we use a lot of AI to develop these solutions and to provide AI capabilities to our clients. So I think those are the three or four pillars of our company.
Brandon: Let’s speak to the broker space. We can talk about captives too, but brokers are probably the bulk of our listeners. They have a tech stack that includes Salesforce or some other CRM, and a client management system, which is usually Applied Epic or Vertafore’s AMS. Add all of this up, and they have a substantial tech stack, right? From a cost standpoint, that tech stack might be 25% of their net profitability at the end of the day, or maybe 2 to 5% of gross sales. But when you look at it from a margin compression standpoint, it can be substantial. So when you engage with a broker, are you trying to holistically integrate most of their entire tech stack? Or are you going after a piece of it, like Vertafore or Applied Epic does?
Illia: We can do both. The answer is simple, right? We can do both. But obviously, if we’re talking about price and price sensitivity, brokers would want to lower this cost, because it’s quite significant. And I think it can be done easily.
I can’t say it’s bad software. The companies you mentioned built it, it has been operating in the market for many years, and it has been quite successful. But I think we are at a particular moment where something has to change, and that change will come sooner or later.
The idea behind our company is that we can start with something small, if they have a niche problem. Or, if they want to fully integrate our tech stack into their environment and lower their costs, that can also be done easily. We can do that job. But obviously, it depends on the client’s environment, setup, and many other things.
Brandon: One of the biggest complaints I personally hear about legacy software systems is the friction each task puts on the individual. Whether it’s filing emails, checking in policies, getting a certificate of insurance, or whatever it is, there are the keystrokes, the screens that open, everything else. There’s all this friction and commitment that comes with doing that. Where do you see yourself in that arc?
Illia: Okay. In terms of how those annoying manual tasks can be solved, I can tell you how it works in the market and in our software. The market is currently leaning toward using AI to solve those problems, obviously. I can’t speak for all companies like us, but I see a trend: they are implementing MCP servers. That means you can use any existing AI provider, like Anthropic or OpenAI, and connect it to the system, and to other systems like Outlook, Gmail, or any other software, through the MCP server. In this case, you work with this traditional or proprietary software through those AI providers and enter data from them. So basically you can just tell ChatGPT: please read my email, extract the data, and create a new request, or add a new client to my system, or potentially even issue a new policy, and many other things, right?
This is what we are doing. We are creating this kind of AI environment inside our solution. So instead of opening ChatGPT or Claude, users open our solution, set up agents, and chat with the system inside our system. Of course, as a company, we use Anthropic, OpenAI, and companies like them to help us with this AI infrastructure. But the idea behind our software is that users will work in our environment, not in a third-party solution.
Brandon: I guess an interesting analogy might be what Meta just released last week with Muse, on a personal level. For those of you who haven’t seen it yet, Meta released Muse, and they actually had a big product release yesterday with Muse and the Muse Charm. Muse is effectively an AI agent: you create a virtual assistant that you give a prompt or a command to, whether it’s “go book my travel tickets” or “send an email to this person until you get an answer.” So effectively, what you’re talking about is maybe an enterprise version of something like that. Would that be an appropriate way to think about it?
Illia: Yes. You know, what Meta is doing, I’ve already been using for some time with OpenAI. I connected my OpenAI account to Outlook and a few other systems, like SharePoint and Microsoft Teams. Just recently, I was looking for a small workspace in one of the cities. I told ChatGPT: please find the top companies. Then I said: these are my requirements, please send them an email. And it just did it. I didn’t even type anything. I just talked to it, and I found a space in maybe two or three hours. So in some regard, this is already in place.
But coming back to our software: you’re right, it’s kind of an enterprise-level version. The thing is that the cost of failure in such software is much bigger than if your personal AI agent fails, right? If you issue a policy with a different coverage, that might be a bad thing, right? So obviously you have to build proper guardrails, and you have to build an audit trail of every single operation, including AI: what prompt the client used, what the system added to it, and many other things. That way, you can trace back, find where an error occurred, and fix it. So yes, the answer is yes: it’s like that, but enterprise-grade.
Brandon: Got it. Before I let you go, let’s talk about where your successes have been. I know you can’t talk about individual clients, but let’s talk about the universe you’ve been working in.
Illia: We have different segments. In the alternative risk segment, our biggest success is in the United States. We work across quite a few domiciles: a lot in Vermont, Delaware, and Utah. We also have some international clients, in Bermuda and the Cayman Islands, and some in Europe.
When it comes to brokers, we work with quite large international brokers. They are worldwide, with representatives in the UK, obviously in the United States, and also in the Middle East. We have some clients in the Middle East. And in terms of traditional insurance companies, we also work a lot in New York. It’s a slower-moving market, but we have some clients there too.
Brandon: Where can people find you? Where can they get hold of you and learn more about DICEUS and how they can get engaged?
Illia: We normally go to many conferences. Last year, I think we visited about 12 conferences in different places. And we get quite solid inbound leads from OpenAI, from ChatGPT, from Claude. People go there, and it seems the level of trust in those platforms is high, and they find us.
The good thing about finding companies through AI is that it’s different from Google search. With Google search, people somehow found ways to get to the top of the results, maybe even without being a professional company. With OpenAI, it’s very difficult to be at the top of the list, because you have to be present everywhere and provide use cases. It actually searches across the internet and finds a lot before providing an answer to users. So I understand why people go there to find information, because I do it too, and the answers are great. I can’t say they’re always very precise, but they’re sometimes better than if I go to Google. And obviously, podcasts like yours also help us a lot.
Brandon: Great. Well, we appreciate you being on the podcast, Illia. Check out Illia Pinchuk and DICEUS. What’s the website, and what’s a contact email people can reach out to you at?
Illia: You can email me directly.
Brandon: Great. Well, thanks so much for being on RiskCellar, and we’ll look forward to your huge inflow of leads coming from this podcast. So thanks so much.
