The insurance industry has a longstanding reputation for meticulous risk assessment, but the integration of artificial intelligence (AI) and third-party data is rapidly changing how insurers quantify and manage risk. On a recent episode of the Unstructured Unlocked podcast, Matthew Grant of InsTech, a seasoned expert with over 25 years in catastrophe modeling, shared his insights on how these advancements are reshaping the industry with podcast hosts Michelle Gouveia, VP at Sandbox Insurtech Ventures and Tom Wilde, CEO of Indico Data.
From leveraging satellite imagery to optimize underwriting to employing AI for more accurate catastrophe models, Grant explored the applications and challenges of using emerging technologies in insurance. Here’s a closer look at the discussion, his perspectives on third-party data, and how the industry can address key hurdles in adoption.
Listen to the full podcast episode here: Unstructured Unlocked Podcast
Unlocking the power of data in property risk assessment
“For someone outside the industry, property insurance may seem simple,” Grant remarked. “How hard can it be to understand where a building is, what it’s made of, and its proximity to hazards like flooding? But in reality, it’s still incredibly difficult for insurers to get that information.”
This challenge underscores the importance of third-party data. By leveraging resources like satellite imagery, lidar, and ground sensors, carriers can gain new insights about buildings and the surrounding environment. For example:
- Satellite imagery provides a bird’s-eye view of properties and nearby hazards like flood zones or wildfires.
- Lidar technology delivers high-resolution 3D maps, enabling precise assessment of building structures and materials.
- Ground-level cameras and sensors help carriers obtain real-time data on environmental risks.
But as Grant pointed out, the key is not merely accessing this data but ensuring its quality, affordability, and reliability. “For insurers to adopt these new data types, they need to trust that the information is better than their existing sources and cost-effective enough to justify investment.”
The rise of AI in catastrophe modeling
AI is playing an increasingly prominent role in catastrophe modeling, providing insurers with greater precision in analyzing risk. For example, in areas like hurricanes and severe convective storms, AI has enhanced risk assessment by addressing the shortcomings of traditional models.
Grant highlighted Zurich Insurance as a case study of innovation, saying, “They’ve built a terrorism model using AI, and their ability to leverage generative AI has been a significant competitive advantage.” By integrating AI into their modeling, Zurich is one of many companies showcasing how cutting-edge technology can transform underwriting and risk evaluation.
Yet Grant reminded us that while models are more advanced today, some uncertainties remain insurmountable. “The short-term unpredictability of events like hurricanes is a massive challenge. Even with 30 years of hurricane modeling, we’re still far from reducing uncertainty entirely.”
Related Content: Guidewire’s Laura Drabik on transforming underwriting with AI and smarter decision-making
Incorporating AI and data into underwriting
The fusion of AI and third-party data isn’t just revolutionizing modeling; it’s also transforming underwriting. According to Grant, insurers often face two significant challenges when integrating these technologies:
- Data hygiene: Insurers need to extract and organize legacy data efficiently across silos. “The data is often there,” Grant explained, “but underwriters can’t always access it easily.”
- Confidence in new data: Adopting cutting-edge datasets requires trust in their accuracy and applicability. Insurers must ensure that incorporating AI-derived insights aligns with their existing underwriting methods.
Grant noted that generative AI is a major driver of efficiency in tackling these challenges. Whether it’s summarizing data or generating recommendations, AI enables faster, smarter decision-making. “The goal should be to use AI not just for incremental gains but for driving tangible improvements in efficiency and accuracy,” he said.
Will insurers build or buy AI solutions?
An ongoing debate in the industry is whether companies should build proprietary AI models or integrate existing third-party solutions. Grant believes the answer varies by organization, explaining, “Some insurers see building their own models as a competitive advantage, while others look to integrate best-in-class tools.”
This decision often depends on a company’s position within the technology adoption curve. “Some are early adopters, leveraging AI tools to gain a competitive edge, while others take a more cautious approach, preferring to use proven, commercially available solutions,” Grant observed.
For many insurers, a hybrid approach may be most effective, balancing in-house innovation with partnerships. Companies that choose to build their own models, for example, can later adopt third-party solutions as these technologies mature.
Barriers to adoption and regulation’s role in AI
One of the biggest hurdles in leveraging AI and data analytics within insurance is navigating regulatory requirements. For instance, in regions like California and Florida, insurers must comply with extensive model approval processes before adopting new tools. “The regulators play a critical role,” Grant emphasized, “but this can sometimes slow down the integration of innovative technologies.”
He added that some insurers are also constrained by internal barriers, such as outdated processes or hesitancy to entrust AI with high-stakes decisions. To overcome these challenges, companies must prioritize transparency, governance, and alignment with regulators.
AI’s potential role in loss prevention
Beyond catastrophe modeling and underwriting, Grant believes AI has an untapped opportunity to drive loss prevention. “There’s always been a tension between insurers covering losses and helping clients prevent them,” he noted. Examples of prevention efforts, such as providing credits for hurricane-resistant roofs, show promise but remain underutilized.
Grant shared that factors like miscommunication, execution challenges, and client engagement have limited progress in this area. However, as data accuracy improves and confidence in AI grows, he foresees a future where loss prevention will play a larger role in insurance practices.
Related Content: Maximizing underwriting profitability through AI-powered document processing
The future of risk modeling in insurance
When asked about the future of risk modeling, Grant expressed optimism about the growing ecosystem of solutions. “We’ve gone from two dominant players in catastrophe modeling to a vibrant space filled with startups offering value around the margins,” he said. “The competition is far less aggressive, creating a collaborative environment where companies can innovate together.”
Grant also highlighted generative AI as a game-changer, noting that its ability to process vast datasets opens up new possibilities for insurers. However, he tempered his enthusiasm with pragmatism, reminding listeners that uncertainty will always be a part of risk assessment.
“Even with advanced models, nature is unpredictable,” he concluded. “It’s essential to balance technological capabilities with underwriting expertise to make decisions in the face of uncertainty.”
Keep exploring AI and data trends in insurance
AI and third-party data are revolutionizing the insurance industry, offering powerful tools for more accurate risk modeling and underwriting. By addressing barriers to adoption and building trust in emerging technologies, companies can unlock significant benefits for themselves and their clients.
Subscribe to the Unstructured Unlocked podcast to get the latest episodes on your favorite platforms, including:
Subscribe to our LinkedIn newsletter.
Frequently asked questions
- How do insurers ensure the quality of third-party data before using it in underwriting? They validate it by comparing it against internal claims data, running pilot programs, and assessing consistency across different sources. Data providers are also vetted for accuracy, completeness, and how well their data integrates with existing systems.
- What specific benefits does generative AI offer insurers beyond summarizing data? Generative AI can simulate risk scenarios, generate customized policy recommendations, automate document processing, and assist underwriters in making complex decisions faster by highlighting key risk indicators from unstructured data.
- Why is loss prevention still underutilized despite the available technology? Because it often requires coordination across departments, ongoing client engagement, and investment in communication tools. Many insurers still prioritize claims response over proactive prevention, and clients may resist or ignore suggested improvements unless incentivized.
