THANK YOU FOR SUBSCRIBING
Insurance CIO Outlook Weekly Brief
Be first to read the latest tech news, Industry Leader's Insights, and CIO interviews of medium and large enterprises exclusively from Insurance CIO Outlook
A featured contribution from Leadership Perspectives, a curated forum for finance technology leaders, nominated by our subscribers and vetted by the Insurance CIO Outlook Editorial Board.
Sam Evans, Founding Partner, Eos Venture Partners, is a venture capital leader focused on insurtech and financial services innovation. Through strategic investments and industry expertise, he supports emerging companies in scaling transformative technologies while helping shape the future of insurance and risk management.
We have seen a clear shift in the market, with both the adoption and funding of AI-native companies increasing rapidly. Furthermore, the way insurers are utilising AI has changed, moving from limited testing around the edges to widespread adoption at the heart of the insurance value chain.
Foundational models and agentic AI are driving the change from earlier technology that offered incremental productivity gains to true operational transformation. Agentic AI represents a fundamental shift from systems that simply generate text or support image recognition to autonomous agents capable of executing multi-step end-to-end workflows.
The Gallagher InsurTech report for Q1 2026 found that 95% of funding flowed to AI-focused businesses. Even allowing for a generous interpretation of what constitutes a true AI business, with many jumping on the bandwagon, the direction is clear. We are seeing a concentration of funding in AI-native companies solving specific, complex workflows.
A data challenge transformed into a competitive advantage
The insurance industry has always struggled with messy, unstructured data that has often been a bottleneck to true automation. Previous attempts to solve this issue have only been partially successful. For example, whilst OCR can extract text with high levels of accuracy, it can lack contextual understanding, breaking down when formats change, or data is presented in non-standard ways. Similarly, RPA requires rigid, rules-based templates that often cannot handle the exception-heavy nature of insurance documents.
The advent of Large Language Models (LLMs) and agentic AI will finally solve this decades-old constraint. Agentic AI systems possess the contextual reasoning necessary to understand a document's meaning, regardless of its format. They act as intelligent translation layers, bridging the gap between unstructured inputs and structured core systems.
We believe that the companies able to create a long-term, sustainable value proposition will be those combining the power of LLMs and agentic AI with structured proprietary data. Ultimately, it will be access to the data (and the insights derived from that data) that creates true differentiation.
Underwriting Reimagined
Historically, underwriters have spent the majority of their time on manual data extraction, triage, and administrative tasks, leaving less time for actual risk assessment.
“The last 18 months have seen leading insurers transition from exploration to enterprise-wide adoption of AI.”
Agentic AI inverts this extraction-to-judgment ratio. Modern AI agents can autonomously ingest submissions, extract relevant data points from hundreds of pages of unstructured documents, perform automated gap analysis, and check against carrier appetite guidelines. The system then presents the underwriter with a synthesized, structured risk profile.
Human in the loop remains fundamental to a successful outcome; AI is not replacing underwriters, rather freeing up their time to focus on high-value decisions. Current projections suggest that AI adoption in underwriting will increase from 14% today to 70% by 20281.
Increased speed and improved accuracy in claims
The claims environment is similarly ripe for agentic transformation. For simple, high-frequency claims, AI agents enable touchless processing. They can automatically extract data from First Notice of Loss (FNOL) submissions, verify coverage, assess damage from images, and calculate standard settlements. For more complex claims, agents provide real-time triage and intelligence, routing the claim to the appropriate specialist with a comprehensive summary already prepared.
Furthermore, agentic AI significantly enhances fraud detection capabilities. By analyzing vast amounts of unstructured data across multiple claims simultaneously, AI systems can identify subtle patterns and anomalies that traditional methods may miss.
A time to embrace change
The operational case for agentic AI is proven, and the mandate is clear: move from exploration to enterprise-wide integration.
That is not to say that the process will be easy. Success involves more than just deploying new software. Enterprisewide adoption of agentic AI requires significant cultural change, training, access to different talent pools, modern data infrastructure and robust governance frameworks. We would also caution against a natural tendency to build in-house, and instead recommend a balanced approach that includes strategic partnerships with early-stage, workflow-specific AI companies, which are likely to improve speed and ensure access to the latest technological advancements.
We believe now is the time for committed action. It will be almost impossible to catch the leading players who can successfully make this transformation, as the benefits to both the business and the customer will be exponential rather than incremental.
