Our Take
Insurance is finally measuring AI ROI in dollars, not deployments—and the metric that matters is risk selection, not headcount savings.
Why it matters
For years, enterprise AI spending was justified by abstract efficiency claims. Insurance's turn toward quantified underwriting returns signals that boards are demanding proof, forcing the entire sector to stop counting use cases and start counting profit.
Do this week
Chief Risk Officer: audit your top 10 AI deployments and map each to claims impact (% fraud caught, premium written at better pricing) before month-end so you can build the business case for your next fund round.
Insurers Are Shifting From AI Ambition to AI Returns
The 2026 Evident AI Index documents a structural reorientation inside the insurance industry. Insurers are no longer competing primarily on the scale of AI investment or the novelty of their deployments. Instead, they are embedding AI directly into underwriting workflows and capital allocation decisions, and they are publishing the financial results.
Three market leaders have led this transparency push. Manulife, Generali, and Intact Financial have disclosed AI-driven value publicly, with projections totaling over $1 billion in value across their respective reporting periods (per the Evident AI Index). This public disclosure of hard returns marks a departure from the earlier phase of enterprise AI, in which companies announced use cases and talent hires without connecting either to bottom-line impact.
Staffing reflects the shift. While the broader insurance workforce contracted 2.2 percent over the past year, AI-specialist headcount expanded 32 percent across 30 tracked insurers. More telling: data engineering roles are declining as a share of the AI talent stack, replaced by roles in AI development and software implementation. AI specialists now represent one in every 50 employees at indexed insurers. Nearly 40 percent of those insurers now assign explicit executive responsibility for AI, with most appointments occurring within the last 12 months.
Deployment patterns are also accelerating toward coordination. Agentic AI systems, which orchestrate actions across multiple stages of the policy administration and claims lifecycle, have surged in adoption. One in four newly disclosed use cases now shows evidence of agentic orchestration, compared to one in twenty six months prior (per the Evident AI Index).
Zurich's Modular Platform Model
Zurich demonstrates one approach to scaling with governance intact. The insurer deployed ZurichIQ, a modular generative AI platform integrated into underwriting, claims, legal, and service operations. The architecture unifies functional tools such as PolicyIQ for contract comparisons and GuidelineIQ for enforcing underwriting standards. A dedicated governance committee manages model risk and AI investment across business lines. Internal training programs, including a £1.3 million AI apprenticeship initiative, reinforce consistent practice. Zurich ranked 4th globally in the Evident AI Index, a jump from 12th position, attributed to this shift toward shared platform over decentralized experimentation.
Claims Costs Drive the Entire Calculus
Claims typically represent 60 to 80 percent of premium income. Minor improvements in fraud detection and risk selection therefore produce disproportionate financial impact compared to general administrative cost reduction. This mathematical reality explains why insurers are redirecting venture capital and internal innovation toward data sources that enable more dynamic analysis of climate volatility and cyber threats. The financial stakes are orders of magnitude higher in underwriting than in back-office efficiency.
Transparent ROI reporting also answers shareholder concerns about AI deployment costs. As boards and investors scrutinize rising AI spending without clear returns, the insurers publishing hard financial results establish the credibility that others will be forced to match. This creates a compliance-like dynamic: once three tier-one players disclose concrete value, the rest of the industry faces pressure to do the same or face questions about why they cannot measure the returns on their own AI investments.
What Underwriting Leaders Should Audit Now
If you are leading underwriting or AI strategy at an insurer, map your existing AI deployments to premium impact, not process speed. Identify which use cases touch the underwriting decision itself, whether through better risk classification, pricing precision, or fraud prevention. Compare their contribution to claims cost against administrative efficiency gains.
Second, examine your governance structure. Zurich's model uses a dedicated committee for AI investment and model risk management. If your organization still treats AI as a technology initiative rather than a business function accountable to underwriting results, you are structurally behind the leaders in the Evident index.
Third, audit your talent composition. If data engineering still dominates your AI hiring, you are investing in foundation work while competitors are hiring for implementation and optimization of business-specific workflows. This signals a delay in moving from infrastructure to outcome.