This blog post is part of a series covering the five pillars of the Intelligent Insurer Operating Model. Click the links here to read the first, second, third, and fourth posts in the series.
I want to address something I hear often in conversations with insurers about the Intelligent Insurer Operating Model for making use of agentic AI: a framing problem. AI transformation in insurance gets described as an IT initiative. Or a business initiative. Sometimes it gets described as “the business telling IT what to build.” None of these framings produce the result organizations are looking for.
I think of it as an orchestra. The business composes the symphony: which workflows to transform, which outcomes to optimize, which friction to remove. The CIO conducts the orchestra to bring that vision to life—the architecture, the guardrails, the delivery engine that allows the business to move fast without creating chaos. A score without a conductor produces noise. A conductor without a score produces nothing. High-impact transformation requires both, in total sync, from the start.
The Infrastructure of Shared Progress
Three enabling conditions make this partnership work in practice.
The first is a shared AI literacy baseline—not technology literacy, but business literacy about what AI can and cannot do. When business and technology leaders operate from different assumptions about AI capability, every initiative generates friction. When they share a common foundation, decisions move faster and with less conflict. I would also argue this is a change management prerequisite: people cannot meaningfully engage with a transformation they do not understand.
The second is a vendor truth layer: a disciplined, jointly owned capability to evaluate vendor claims against your actual operating environment and integration constraints. The market for AI products in insurance is crowded, the claims are often outsized, and the gap between a compelling demo and a production-ready integration is frequently significant. A rigorous evaluation process, owned jointly by business and technology, closes that gap before it becomes a failed implementation and a credibility problem.
The Technical Wiring
On the infrastructure side, two developments merit particular attention.
Model Context Protocol (MCP) is an open standard that allows AI models to connect to legacy policy and claims systems without custom-coding every integration. I describe it as the USB-C of the AI world: a standardized connection that works across systems, reducing the integration overhead that has historically slowed AI deployment in complex enterprise environments.
OpenClaw is an open-source toolkit that gives digital employees the ability to interact with existing insurance software (e.g., underwriting workbenches and claims systems) the way a human would, but at machine scale and precision. If MCP is the wiring, OpenClaw or for that matter any “-Claw” is what allows the digital workforce to act on what it finds.
Neither is a silver bullet. But both represent the kind of standardized infrastructure that allows AI capability to scale across an enterprise without requiring bespoke engineering for every new application—which matters enormously when you are trying to build at speed.
Where to Start
My recommendation: Host a 60-minute “Shared Literacy” session where Business and IT leaders define what “Agentic AI” means for your specific company. Eliminate the jargon gap before you approve the next budget.
Discuss the Intelligent Insurer Operating Model in person! Join us on May 21st in Chicago at our Regional Property Casualty Insurance Forum on Agentic AI or on June 16th in Des Moines for our Regional Life Insurance, Annuities, and Group Benefits Forum on Agentic AI. Designed for senior technology, operations, strategy, data, analytics, and line of business leaders, these forums explore how AI-driven intelligence is being embedded across underwriting, new business, policy administration, claims, enrollment, and customer servicing and feature numerous innovative service providers implementing agentic AI in production.
Next: Pillar Five—the strategic case for speed, and why the organizations that master this model are not just more efficient, they are structurally different.
