Datos Insights Flags Agentic AI Reshaping Insurance Core Systems

IIR reports that Mitch Wein, Head of Insurance Strategy and Advisory at Datos Insights, described early signs that agentic AI is challenging where underwriting, policy administration and other insurance core functions should live. IIR quotes Wein saying, "The core system vendors are probably in trouble." IIR also quotes Wein describing prior agentic limits: "You couldn't string together the workflows, You couldn't move stuff from one LLM to another LLM or SLM. It would lose context. The data would get exposed. It wouldn't retain the rules about the data or the lineage if it was transformed." Reporting by IIR notes that Anthropic introduced MCP in November 2024, and that Anthropic donated MCP to the Agentic AI Foundation under the Linux Foundation in December 2025.

Industry-pattern observations: agentic AI architectures depend on chaining models, data sources and business logic while preserving context and governance. Standards like MCP are being positioned in public reporting as a mechanism to provide secure, two-way connections and context propagation across agents, services and data stores. For practitioners, that pattern raises operational questions about interoperability, access control enforcement, and end-to-end lineage across model handoffs.

Industry observers note that regulated sectors such as insurance place additional requirements on data handling because carriers process non-public personal information and health-related data. Public reporting frames MCP as addressing some of those concerns by enabling access controls and provenance as workflows traverse multiple systems and model endpoints. Vendors and integrators that support cross-system orchestration are likely to become focal points for implementation discussions, according to the coverage.

Indicators an observer might follow include vendor support for MCP connectors, reference implementations from major platform providers, early insurer pilots that demonstrate cross-system lineage and governance, and contributions to the Agentic AI Foundation repository. Stakeholders should monitor technical documentation and pilot results to evaluate how well MCP preserves access controls and data lineage in multi-agent insurance workflows.