The END of Slow Insurance? AI Underwriting That's Changing the Industry
In this episode of The Curiosity Code podcast, host Alex Khomyakov chats with Parul Kaul Green, former Chief Digital Strategy Officer at Liberty Specialty Markets and a pioneering voice in AI-driven insurance innovation. With leadership roles at AXA and Aviva, Parul has spearheaded digital underwriting platforms and AI-powered claims transformation. Now advising insurtech startups, she focuses on integrating AI into complex, regulated insurance environments without sacrificing profitability or compliance. The discussion dives into AI’s evolving role in underwriting and claims—highlighting how AI agents streamline submission processing, reduce fraud, and improve risk selection. Parul also addresses the trade-offs between bolt-on AI tools and full digital transformations, stressing the importance of data quality, governance, and human-in-the-loop protocols. She provides strategic advice for insurtech founders, sharing lessons from global markets like Asia, the U.S., and Europe. From NLP-powered underwriting to agentic AI and IoT-enhanced risk modeling, Parul predicts a future where insurers gain competitive advantage through smarter, ethical, and scalable AI deployments. This episode offers essential insights for anyone navigating AI’s transformative impact on insurance.

Alex: Hey everybody and welcome to another episode of The Curiosity Code podcast. I’m Alex Khomyakov, your host, and today I’m joined by an incredible guest — Parul Kaul Green, former Chief Digital Strategy Officer at Liberty Specialty Markets and one of the most insightful voices in AI-led insurance transformation. Parul has held leadership roles at AXA, Aviva, and other top-tier insurers and now works with insurtechs and boards driving ethical, scalable innovation. Parul, welcome!

Parul: Thank you, Alex! Really excited to be here and talk about the intersection of AI, insurance, and transformation—it’s one of my favorite topics.

Alex: So let’s start with your journey. You’ve held executive roles in large insurance companies, led innovation programs, and now you’re advising startups. What pulled you into the AI space?

Parul: I’ve always been fascinated by data and decision-making. Early in my career, I realized that much of what we do in insurance—from underwriting to claims—is fundamentally about information. When AI started to mature beyond hype, I saw its potential to transform those processes. My work evolved from digital strategy to building AI capabilities into underwriting, pricing, and customer engagement.

Parul: But I also became aware of the risks—bias, explainability, governance—and how quickly things can go wrong without proper oversight. That’s what pushed me to focus not just on deploying AI but doing so responsibly.

Alex: Love that framing. Let’s get into specifics. In underwriting, where are you seeing the most exciting use cases for AI right now?

Parul: So underwriting is evolving in two directions: augmentation and automation. On the augmentation side, AI helps underwriters process submissions faster—extracting information from documents, summarizing risk factors, and benchmarking against historical data. This saves time and improves decision-making.

Parul: On the automation side, we’re seeing straight-through processing for simple risks—especially in SME or parametric insurance. AI models trained on historical bind/decline data can make instant decisions with high accuracy. But the key is knowing when to let humans step in. High complexity or high-value cases still need expert review, and that’s where hybrid models shine.

Alex: That’s super clear. What about claims? What’s changing there?

Parul: Claims is a goldmine for AI. The biggest shift is around triage—automatically categorizing claims by complexity, risk of fraud, and urgency. NLP models can read claim descriptions and route them accordingly. Image recognition is being used in auto and property to assess damage, sometimes even assigning a repair estimate automatically.

Parul: Fraud detection is another huge area. AI models are excellent at spotting anomalies—patterns that suggest suspicious behavior across claims, customers, or even service providers. And as insurers get more sophisticated, these models are becoming more nuanced, detecting fraud earlier and reducing false positives.

Alex: You mentioned governance earlier. Let’s dive into that. What does responsible AI look like in insurance?

Parul: Responsible AI in insurance means aligning model development and deployment with ethical principles, business goals, and regulatory standards. That includes fairness—ensuring AI doesn’t discriminate against certain groups; explainability—being able to justify a model’s decision; and governance—clear accountability over how models are trained, monitored, and updated.

Parul: One mistake companies make is thinking AI is just a tech issue. It’s not. It’s a business risk and needs cross-functional governance. I always advise creating an AI ethics committee or oversight framework that includes underwriting, legal, compliance, and tech leaders working together.

Alex: That’s powerful. What are the trade-offs when deciding whether to build AI tools in-house versus partnering with insurtechs?

Parul: It depends on your speed, expertise, and risk appetite. In-house gives you control and can be deeply integrated, but it’s expensive and time-consuming. Partnering with insurtechs lets you move faster and tap into specialized capabilities, but you need strong APIs and data standards to integrate well.

Parul: Also, the build vs. buy decision should consider explainability and oversight. If you buy a black-box model without knowing how it works, you take on a big regulatory and reputational risk. Transparency and alignment are key.

Alex: You’ve worked across markets—Europe, Asia, the U.S. What are some interesting regional differences you’ve seen in AI adoption?

Parul: Great question. Asia is moving incredibly fast—markets like Singapore and Hong Kong have supportive regulators and a strong tech talent pool. U.S. insurtechs are more aggressive with experimentation, especially in embedded insurance and on-demand products. Europe is more cautious but is leading in AI governance, especially with the upcoming AI Act. So you see a spectrum—speed in Asia, creativity in the U.S., and policy maturity in Europe.

Alex: Fascinating. Let’s talk about agentic AI and LLMs. Are you seeing these being used yet in underwriting or claims?

Parul: Absolutely. LLMs are changing how we interact with data. In underwriting, they can summarize submissions, extract key clauses, and even draft emails or coverage explanations. In claims, they’re being tested for generating case summaries or pulling precedent data. The challenge is reliability—LLMs can hallucinate, so you need guardrails. But when combined with structured data and expert feedback loops, they’re a powerful augmentation tool.

Alex: So exciting. What advice would you give to founders building in insurtech today?

Parul: Solve real problems. Don’t build AI for the sake of it—find inefficiencies or pain points and ask, “Could AI reduce friction here?” Second, work closely with actuaries, underwriters, and claims professionals. They’re your end users. And third, invest in explainability and data quality early. These will make or break adoption.

Alex: Parul, this has been phenomenal. Thank you so much for sharing your insights on the future of AI in insurance!

Parul: Thank you, Alex—it’s been a pleasure!

Alex: And to our listeners—thanks for tuning in. Don’t forget to subscribe and share this episode. Catch you next time on The Curiosity Code.