Why companies that treat AI as a technology upgrade are solving the wrong problem.
By James W. Moore | InsuranceIndustry.AI
Key Takeaways
- AI is not a productivity upgrade. It is a structural disruption to the economic logic that determined how insurance organizations were built in the first place.
- Firms were designed to minimize coordination costs. AI is compressing those costs, turning entire layers of organizational structure into artifacts of a problem that no longer exists at the same scale.
- The distinction between AI as a technology and AI as a business model is the difference between carriers that will lead and carriers that will follow.
- The question is not whether your organization is adopting AI. It is whether your architecture was designed for the world that just ended.
When steel-frame construction arrived in the late 19th century, it didn’t make buildings better. It made different buildings possible. The skyscraper wasn’t an improved brownstone. It was a different structural logic entirely. Architects who tried to retrofit steel frames onto load-bearing masonry walls didn’t get taller buildings. They got structural failures.
Most insurance carriers today are retrofitting.
They are bolting AI onto organizational structures designed for a different economic era, structures built to solve problems that AI is quietly making obsolete. The technology is real. The results are underwhelming. And the reason has nothing to do with the quality of the tools.
The Restructuring Is Already Atmospheric
AI is not arriving. It is already here, and the gap between those who understand what it is and those still debating readiness is widening faster than most executives realize.
The inflection point rarely announces itself. It shows up in a competitor’s expense ratio first.
Nearly nine in ten organizations globally are now regularly using AI, according to McKinsey’s 2025 State of AI report. Yet only 39% report any measurable impact at the enterprise P&L level. In insurance specifically, the Roots 2025 State of AI Adoption report found that more than 90% of carriers are exploring or testing AI, but only 22% have deployed solutions in full production.
That gap is not a technology problem. Carriers are not failing to find capable AI tools. They are failing to build organizations capable of absorbing what those tools make possible.
What AI Actually Changes
To understand what AI is disrupting, it helps to understand what it is replacing.
In 1937, economist Ronald Coase asked a deceptively simple question: why do firms exist at all? His answer, transaction cost theory, shaped a century of organizational thinking. Firms exist, Coase argued, because coordinating activity internally is often cheaper than coordinating it through markets. The cost of searching for information, negotiating agreements, and managing relationships across organizational boundaries is high enough that bringing functions inside the firm makes economic sense.
Douglass North extended this framework by showing that institutions, both formal rules and informal norms, emerge to reduce those coordination costs within a broader societal context. Firms provided a structure to manage complexity and uncertainty more efficiently than a collection of individual market transactions. You built a department because the alternative was chaos.
AI is changing that calculation at its foundation.
The distinction between AI as a technology and AI as a business model is critical. Technology makes existing processes faster. Business model change alters how value itself is created and captured. When AI compresses coordination costs (the costs that justified building the organization the way it was built), the organizational forms designed around those costs don’t automatically update. They persist. And they become expensive.
Consider three specific shifts:
Coordination costs. Before AI, coordinating activity across functions required meetings, handoffs, management layers, and middleware systems. After AI, much of that translation work happens continuously and automatically. Functions that existed primarily to manage information flow between other functions become candidates for consolidation.
Specialized decision-making. Before AI, credentialed expertise was genuinely scarce. An experienced commercial underwriter or a seasoned claims adjuster carried judgment that was rare and expensive precisely because it took years to develop. After AI, model-assisted judgment is broadly accessible at near-zero marginal cost. The scarcity equation is shifting.
The labor-to-decision ratio. Before AI, one complex decision might require five people across two departments and three days of elapsed time. After AI, one person working with capable models can produce decision-grade output in a fraction of that time. That is not a productivity improvement. That is a different team architecture.
What this summary does not fully capture: coordination costs compress, but verification costs remain. The new architecture does not mean fewer people. It means fewer coordinators and more adjudicators.
The best underwriters, adjusters, and analysts don’t disappear. They stop being routers and become referees. That is not a demotion. It is a different and arguably more valuable role.
A February 2026 working paper by economist Alex Farach formalizes this argument. Titled “AI as Coordination-Compressing Capital” and currently an arxiv preprint awaiting peer review, it describes AI systems that reduce coordination costs as expanding managerial spans of control and making previously necessary organizational layers redundant. The uncomfortable implication is that entire departments may turn out to be scaffolding built around coordination friction rather than load-bearing walls. Not because the work wasn’t valuable — it was. But the work existed to solve a coordination problem that AI is now solving differently.
This is why incumbents underreact. The existing hierarchy isn’t just an organizational chart. It encodes the economic logic of a prior era. Asking it to evaluate its own obsolescence is a structural conflict of interest.
The Insurance Case
Insurance makes this argument more clearly than almost any other industry, because insurance was always coordination-intensive by design.
The carrier model, with integrated underwriting, claims, actuarial, distribution, and compliance under one roof, was built precisely because the coordination costs of managing those functions separately were prohibitive. Bringing them inside the firm made economic sense when information was expensive, expertise was scarce, and managing relationships across organizational boundaries required armies of people.
Consider what that means operationally. A commercial submission that once required three underwriters and 72 hours to clear can now be triaged, enriched, and routed by one underwriter working with AI in under an hour. The other two underwriters did not get faster. They became unnecessary for that task. That is not an efficiency gain. That is a structural change disguised as a workflow improvement.
KPMG’s 2025 Insurance CEO Outlook, which surveyed 110 insurance CEOs globally, captures the broader tension: 73% are prioritizing AI investment to streamline underwriting, claims, and customer experience. At the same time, 77% identify workforce transformation and upskilling as a top constraint on growth. Those two numbers are not in tension. They are cause and effect.
Insurers are not struggling to find AI tools. They are struggling because, as Insurance Thought Leadership observed in its 2026 outlook, after a decade of patching legacy platforms and layering point solutions, what is now required is not another upgrade cycle. It is a structural re-architecture of how the business operates.
Automation bolted on top of a fragmented architecture cannot scale sustainably. The structure has to change, not just the tools inside it.
The strategic path forward for carriers willing to make that distinction is developed fully in our recent white paper, The AI Sidecar Strategy. The short version: you don’t have to replace everything. But you do have to be honest about which parts of your organization exist because of genuine business logic and which parts exist because coordination used to be expensive.
Insurance Is the Canary
Insurance is not a special case. It is the clearest case.
Banking, healthcare, professional services — any industry built around integrated operational complexity faces the same structural question. But insurance makes it visible because the coordination costs were always so high and so explicit. The org chart of a major carrier is a map of what it once cost to move information and expertise through the enterprise.
The rule of thumb worth applying across industries: any business where margin comes from proprietary coordination rather than proprietary risk is exposed. Insurance executives will recognize that distinction immediately.
And the competitive threat is not abstract. AI-native carriers entering the market may not look like smaller versions of incumbent carriers. They may look like entirely different organizational species, built from scratch around the new economics rather than adapted from the old ones.
BCG’s most recent research on the future of corporate functions confirms what leading carriers are already experiencing: corporate functions are becoming leaner, faster, and more autonomous as AI-driven workflows reduce costs, simplify leadership structures, and reshape how performance is measured. This is not a prediction. It is a description of what is already happening at the industry’s leading edge.
The carriers pulling ahead are not doing so by running better pilots or deploying more tools. They are doing so by asking a harder question about what their organization is actually for, and rebuilding around the answer.
The winners won’t be the ones who automate fastest. They will be the ones who reorganize earliest around the new economics.
The Question Worth Asking
There is a version of the AI conversation that stays safely inside the boundaries of technology adoption. Which tools to buy. How to govern them. How to train the workforce to use them responsibly. These are legitimate questions, and they deserve serious answers.
But they are not the most important question.
The most important question is whether your organization was designed for a world in which coordination was expensive, expertise was scarce, and managing complexity required layers of structure, and whether you are prepared to examine what changes now that all three of those assumptions are under pressure simultaneously.
Not everything built around coordination costs is obsolete. Underwriting judgment, claims relationships, and regulatory expertise: these are not scaffolding. They are the load-bearing walls of what insurance does. But the scaffolding built to coordinate those functions? That is worth examining hard.
Every organization built around the economics AI is replacing should be asking one question: Was our architecture designed for the world that just ended?
The AI Sidecar Strategy, our most recent white paper, develops a practical framework for carriers navigating this transition. Download it free here.
Sources
- McKinsey/QuantumBlack, “The State of AI in 2025”, November 2025
- Roots.ai, “2025 State of AI Adoption in Insurance”, December 2025
- California Management Review, “From Coase to AI Agents: Why the Economics of the Firm Still Matters in the Age of Automation”, April 2025
- Farach, “AI as Coordination-Compressing Capital”, arxiv preprint, February 2026
- BCG, “Making AI Productivity Deliver Real Value”, May 2026
- KPMG, “2025 Insurance CEO Outlook”, January 2026
- Insurance Thought Leadership, “2025 Reflections & 2026 Outlook”, December 2025
- BCG, “Corporate Functions of the Future”, May 2026
AI Disclaimer: This content was created with assistance from artificial intelligence technology. While content is based on factual information from the source material, readers should verify all details directly with the respective sources before making business decisions.
