Nikolai Braiden is a seasoned expert in the FinTech space and a prominent early adopter of blockchain technology who has spent years advocating for the digital transformation of financial systems. With extensive experience advising high-growth startups, he specializes in leveraging cutting-edge technology to drive innovation in lending and insurance underwriting. As the property and casualty sector faces unprecedented challenges from climate-related disasters, Braiden’s perspective on the intersection of data science and risk assessment has become a vital resource for the industry. In this conversation, we explore how carriers are navigating the aftermath of the devastating 2025 Los Angeles wildfires and the strategic role that AI-driven models play in restoring market capacity.
The January 2025 Los Angeles wildfires were a watershed moment for the industry, resulting in $40 billion in insured losses and the destruction of over 16,200 structures. In light of such massive devastation, how are insurers finding the confidence to re-enter or expand their presence in the California homeowners market?
The shockwave from the Eaton and Palisades fires was felt across the global insurance community, with the Swiss Re Institute marking it as the largest insured wildfire loss event ever recorded. However, the confidence we see today from carriers like DUAL North America, Kingstone, and Windward Risk Managers stems from a shift toward disciplined, data-driven precision rather than broad market avoidance. These insurers are moving past the “black box” approach of the past and are now using models that allow them to view risk property by property. By understanding that wildfire exposure is not uniform across the state—and that only 11% of California homes actually sit in the highest risk tiers—they can identify pockets of safety even in regions that were previously deemed uninsurable. This ability to differentiate risk within the same territory is exactly what allows capacity to stay in the market when others are heading for the exits.
We are seeing companies like Windward Risk Managers transition from the hurricane-heavy Florida market into California’s wildfire zones; what does this tell us about the universal application of AI in catastrophe modeling?
It demonstrates that the core challenge of modern insurance is the same regardless of the peril: the need for granular, parcel-level intelligence. Windward is a perfect example of a firm that already relies on property-specific data to assess structural characteristics in hurricane-prone Florida, and they are now simply applying that same rigorous methodology to wildfire risk in the West. This isn’t a blind expansion; it is a strategic move underpinned by models that analyze over 2,000 historical wildfire events and use machine learning to weigh factors like topography and building materials. When a specialty program administrator like DUAL expands their use of AI from wind and hail underwriting into the wildfire space, they are proving that a consistent, technology-first approach is the only way to build a sustainable portfolio in 2026. This level of detail allows them to offer coverage with a clarity and consistency that traditional underwriting simply cannot match.
ZestyAI’s research found that their model classified 94% of the Palisades burn area as high risk before the fire even happened. How does this kind of predictive accuracy change the conversation between insurers and the regulators who approve their rate filings?
That level of precision is a total game-changer for regulatory relations because it replaces speculation with verifiable, data-backed evidence. When a model can accurately flag 87% of the Eaton burn area and 94% of the Palisades area as high risk before a spark is even flying, it proves to regulators that the insurer isn’t just hiking rates arbitrarily. This is why we’ve seen over 200 regulatory approvals for these types of AI models across the U.S., including being the first of their kind approved for carrier rate filings in California. For a company like Kingstone, which is entering California on an excess and surplus lines basis, this accuracy serves as the foundation for their rating and accumulation management. Regulators are more likely to support these new programs when they see that the technology can specifically identify the small percentage of truly high-risk homes while keeping insurance accessible for the majority of the population.
What is your forecast for the evolution of AI-driven risk models in the insurance industry over the next few years?
By 2028, I expect that property-level AI modeling will be the standard requirement for any carrier looking to maintain a license in high-volatility regions. We will move beyond just assessing “fire” or “wind” and start seeing truly integrated models that provide a real-time “resilience score” for every structure, updated as homeowners make changes to their vegetation or building materials. The precision we are seeing now in California—where only 11% of the state is flagged as high risk despite the massive 2025 losses—will become even more refined as more sensory and climate data is ingested. Ultimately, insurance will transform from a reactive industry that pays for damage into a proactive partner that uses these models to help homeowners actively mitigate their risks and lower their premiums.