As a leading technologist specializing in the real-world application of artificial intelligence, Laurent Giraid has a unique vantage point on how machine learning and natural language processing are fundamentally reshaping legacy industries. He joins us today to dissect the profound strategic and operational shifts occurring within the insurance sector. We’ll explore the delicate balance between augmenting human expertise and driving automation, the transition from foundational innovation to an AI-powered future, and the concrete results this technological revolution is delivering—from dramatically faster underwriting to the complete reimagining of claims processing.
You’ve emphasized that human expertise remains a core competitive advantage, even with over 20,000 professionals using AI. How do you balance empowering employees with AI while ensuring their unique skills drive growth? Could you share a specific example of this synergy in action?
That’s the essential question, isn’t it? The most successful strategies don’t see AI as a replacement for human intellect but as a powerful amplifier for it. It’s about taking the mundane, the repetitive, and the data-intensive tasks off an expert’s plate so they can focus on the nuanced, high-level judgment that truly drives value. Take commercial underwriting, for instance. An underwriter’s real skill isn’t in sifting through mountains of documents; it’s in synthesizing that information to understand complex risk characteristics. Now, generative AI agents are deployed to mine both internal and external data sources, summarizing past claims and highlighting key risk factors. The underwriter receives a concise, actionable summary, allowing them to apply their experience to the most critical part of the job: making the final, expert decision. This synergy transforms their role from data-gatherer to strategic risk analyst, which is exactly where their unique value lies.
With the claims call center workforce reduced by a third due to automation, how have you managed this transition for your employees? What steps are you taking to upskill your team to collaborate effectively with new generative AI voice agents and digital processing tools?
That kind of workforce reduction is a significant and sensitive transformation. The key is a proactive, two-pronged approach. First, you streamline operations; in this case, consolidating four call centers into two reflects the new reality of call volumes. Second, and more importantly, you redefine the roles of the remaining team members. They are no longer just handling initial intake calls. Instead, they are becoming the specialists who manage the escalations and the complex cases that the AI cannot—and should not—handle alone. The upskilling focuses on teaching them to be effective collaborators with the technology. This means understanding how the generative AI voice agent works, when to intervene, and how to interpret the data from digitally processed claims to provide a higher level of customer service for those who still prefer a human touch. It’s a shift from being a processor to a problem-solver.
You’ve mentioned a strategic move from “Innovation 1.0” to an “Innovation 2.0” powered by AI. Beyond efficiency, what new capabilities or customer experiences does this unlock? Could you walk us through the most significant operational change this new phase has introduced so far?
“Innovation 1.0” was about building a solid digital foundation—the essential infrastructure and skill sets. “Innovation 2.0” is where that foundation is supercharged by AI to create entirely new ways of operating. The biggest unlock beyond pure speed is the ability to deliver hyper-personalized and proactive service. For example, instead of just reacting to claims, AI can help predict risk and inform product development in near real-time. But the most significant operational change we’ve seen is the mainstreaming of straight-through processing. The fact that over half of all claims are now eligible for this fully automated pathway is a monumental shift. It fundamentally changes the customer experience from a potentially days-long process into a seamless, almost instantaneous interaction for a majority of cases. This isn’t just an improvement; it’s a complete redefinition of what a customer can expect from their insurer.
A 30% reduction in handle times for personal insurance underwriting is a major achievement. Can you detail the proprietary AI models that score risk and consolidate data to achieve this? What was the biggest challenge in integrating these advanced tools into the existing underwriting workflow?
That 30% figure is incredibly impressive, and it’s the result of a multi-layered AI approach. It begins with a proprietary AI-enabled predictive model that acts as a first-pass filter. This model scores every single account in the property portfolio, algorithmically identifying which ones have the highest probable risk of loss. This is a game-changer because it immediately directs human attention where it’s most needed. Then, for those flagged accounts, a generative AI tool steps in. It consolidates all the relevant data into a digestible summary for the underwriter. The biggest challenge wasn’t the technology itself, but the human side of integration. You have to build trust. Underwriters have honed their intuition over many years, so you need to demonstrate that these tools are reliable aids, not replacements for their judgment. It requires a period of running the models in parallel, validating the outputs, and showing underwriters how the AI summaries consistently lead to better, faster decisions.
More than half of all claims are now eligible for straight-through processing. What have you learned about customer preferences from this rollout, and what are the key factors that make a claim suitable for full automation versus one that still requires an expert’s touch?
The rollout has been a fascinating lesson in customer behavior. We’ve learned that when given the option for a fast, digital, and self-service path for straightforward claims, customers embrace it enthusiastically—adoption is around two-thirds of cases where it’s offered. They value speed and convenience above all else in those situations. However, we’ve also confirmed that for more complex or emotionally charged claims, people still want the reassurance of a human voice. The key factor for automation suitability is data completeness and low ambiguity. A simple, well-documented auto claim with clear fault is a perfect candidate. But a claim involving significant property damage with multiple parties or a liability dispute has too many variables and requires the empathy and nuanced problem-solving skills of an experienced claims professional.
In specialty insurance, AI has reportedly cut submission intake times from hours to just minutes. Could you describe the specific AI application behind this improvement and explain how it has reshaped the daily responsibilities and focus of the underwriters on that team?
The application at work here is a sophisticated form of intelligent document processing, likely powered by natural language processing and computer vision. Specialty insurance submissions are notoriously complex and unstructured—they come in various formats, full of dense, industry-specific jargon. The AI is trained to read and understand these documents, extracting the critical data points and populating the system automatically. This eliminates hours of manual data entry. The impact on an underwriter’s day is profound. Their morning is no longer a race against a growing pile of paperwork. Instead of spending their time on tedious intake tasks, they can immediately begin the high-value work of risk analysis, pricing, and building relationships with brokers. Their focus shifts entirely from clerical processing to strategic decision-making, which is not only more efficient but also far more engaging and professionally satisfying.
What is your forecast for the role of AI in the property and casualty insurance industry over the next five years?
Over the next five years, I believe AI will become the central nervous system of the P/C industry, moving from a competitive advantage to a foundational requirement. We’ll see agentic AI, which is already embedded in operations at forward-thinking companies, become widespread. These AI agents won’t just analyze data; they’ll execute complex, multi-step workflows, from prospecting new business to managing stakeholder interactions in the claims process. The focus will shift from discrete AI tools that solve one problem to an integrated AI ecosystem that optimizes the entire insurance value chain. This will lead to faster product development, more accurate and dynamic pricing, and a level of customer service that is both hyper-efficient for simple needs and deeply empathetic for complex ones. The companies that thrive will be those that don’t just adopt AI, but fully integrate it into their culture, amplifying their human expertise to deliver outcomes that are simply not possible today.
