Share with your CIO
Insurance is consolidating around a blunt verdict on AI pilots: they are not the destination. Intelligent Insurer’s Agentic and Generative AI for Insurance USA conference, set for New York on November 18-19, is organizing around that premise, pulling together executives from Liberty Mutual, AXA XL, Swiss Re, Hippo, and Zurich North America to address scale rather than experimentation. Evident’s Q4 2025 tracker puts 68% of publicly disclosed insurance AI deployments in the generative or agentic category, with agentic accounting for 21%. The production gap is real, and the industry knows it.
What this means for your business
The stat that should arrest any insurance CIO is the one nobody at this conference is disputing: most carriers can demonstrate a pilot, almost none can point to agentic AI moving their loss ratio or cost-to-serve at scale. That asymmetry between “using AI” and being structurally advantaged by it is where competitive separation actually happens over the next decade, and the CIOs who treat legacy data architecture and systems integration as solvable later are already behind the ones treating them as the primary constraint now.
The conference agenda reflects a pattern that keeps repeating across enterprise AI deployments in regulated industries, which you could call the governance-integration squeeze. Carriers face pressure to move fast on automation, but every agentic deployment, where software agents take sequences of actions across systems with minimal human hand-holding, immediately runs into model validation requirements, explainability obligations from regulators, and the question of who is accountable when an AI-assisted underwriting or claims decision goes wrong. The organizations that resolve this squeeze fastest are not the ones buying the most sophisticated models; they are the ones that built auditable data pipelines and clear human-override protocols before they scaled. The conference is right to center that sequencing.
The ROI framing in the closing session is the one to watch. If your finance and technology teams are still measuring AI contribution through cost savings alone, you are building the wrong business case and you will lose the budget argument to the next downturn. Underwriting quality, risk selection accuracy, and speed-to-market are the metrics that connect AI investment to durable margin, and any vendor or internal team that cannot speak that language in 2026 is selling you the pilot era dressed up as a production strategy. The falsification condition here is straightforward: if your agentic deployments cannot be connected to a named underwriting or claims metric by mid-2026, the architecture is wrong, not just the measurement.
Based on reporting from Moving AI from promise to performance at Agentic and Generative AI for Insurance USA, originally published 2026-08-04 08:10:00.
