P&C insurers turn to agentic AI to orchestrate workflows

ISG’s July 23, 2026 release on its global Insurance Services - Property and Casualty BPO report said P&C insurers are shifting from simple task automation to agentic AI that can orchestrate work across claims, underwriting support, policy servicing and customer operations. The operational test is no longer whether software can answer questions; it is whether it can sequence FNOL intake, claims triage, document handling and approval steps without losing control.

In Stamford, Conn., ISG said insurers are seeking partners to modernize IT, deploy governed AI at scale and ensure local compliance. That language puts the build-vs-buy decision squarely on governance, not novelty. Carriers want technology that reduces handoffs and supports decision-making, but they also need workflow design, data quality and human oversight before agents can be trusted in production.

The report also frames outsourcing as a broader operating-model tool, describing technology outsourcing as a strategic capability that supports business agility, resilience and long-term transformation. That matters in P&C because the pressure points are structural: high volume claims, exception-heavy underwriting, regulatory review and recurring customer service work all create places where a semi-autonomous system can save time only if the underlying process is already well controlled.

ISG’s earlier P&C coverage shows the market has been tracking provider execution for years. The firm’s 2024 North America Insurance Services Quadrant Report and 2024 U.S. Insurance BPO report were already assessing provider performance, and a November 2025 Insurance Services strategic capabilities report extended that framework into the global market. The 2026 report therefore lands less as a concept paper than as a new stage in a long-running vendor selection cycle, where buyers are comparing service providers on process expertise, platform integration and AI orchestration rather than on application features alone.

The governance warning is reinforced outside insurance. IBM said on January 19, 2026 that e& and IBM were deploying “enterprise-grade agentic AI” into governance and compliance systems in Davos, Switzerland, a sign that regulated industries are already treating agentic systems as operational infrastructure. A 2025 Dataiku session on modernizing governance for agentic AI at scale made the risk side explicit, warning that without guardrails agents can produce hallucinated outputs and hidden compliance gaps.

For P&C insurers, that leaves a narrow buying brief: systems that can support FNOL, claims, underwriting and policy servicing, but only with audit trails, exception handling and human-in-the-loop controls built in from the start. The software stack is moving toward orchestration; the operating model now has to move with it.