For most of the last decade, carriers bought their core systems. Building from scratch rarely made sense, and most of them knew it. Then AI coding tools arrived and the math started to look different. A small team can now stand up working software faster than ever before. Gartner projects that 75% of enterprise engineers will use AI code assistants by 2028, up from under 14% in early 2024. Driven in part by AI-focused advisors, the build-vs.-buy question for insurance core systems is back on the table. But was writing the code ever the hard part? The Question That Came Back The renewed interest in building has as much to do with cost pressure as capability. Technology spend as a percentage of direct written premium has climbed steadily over the past decade, reaching 4.6% of DWP in 2026. Cloud costs never settled the way carriers expected after the on-premise-to-cloud shift. Against that backdrop, an internal build might appear to offer greater control over spend without giving up the capabilities carriers depend on. AI coding tools make the build case feel even more viable. Prototypes that once required months of engineering can now be developed in weeks. For lightweight applications such as portals, checklists, and workflow automation, the speed and value have increased while the cost has dropped. Carriers are already building point tools with AI assistance, and for those narrower use cases, the model can make sense. The more important question now in the build-vs.-buy debate is what they are building on top of. AI Lowered the Cost of the Prototype AI has lowered the cost of the prototype. A small team with modern coding tools can produce software that looks polished, passes a demo, and generates genuine confidence internally. But a working prototype and a production system a carrier will operate for the next decade are different things. Prototype costs capture only the initial build. The larger commitment comes after launch though ongoing maintenance, compliance updates, staff continuity, integration with bureaus and carriers, and the ongoing work of keeping the system current as lines of business evolve. The prototype is where AI makes a visible difference. Owning the production system for the next decade is a different proposition. The Costs That Recur Every Year After Build cost is a one-time number. Operating cost is a compounding one. Once a carrier owns a proprietary core system, it bears the cost of keeping engineers who understand the codebase, updating the system as regulatory requirements shift, and integrating with an ecosystem that keeps changing. That creates an added opportunity cost as well. Every engineer maintaining the core system is unavailable for the targeted builds where internal expertise creates real differentiation. Those are the tools, workflows, and automations that reflect a carrier’s specific knowledge and competitive position. This is the cost that tends to be underweighted in the early stages of a build evaluation. The prototype demonstrates that a solution can be built. It doesn’t account for the years of maintenance, investment, and tradeoffs required to keep it viable. Domain Expertise Doesn’t Transfer Through a Code Generator The speed at which AI produces code has outpaced the domain knowledge required to validate it. A carrier building an insurance system of record still has to define the requirements, business rules, exception handling, and regulatory logic the system will enforce, then validate that the code reflects them correctly. Logic errors introduced through AI-generated code can reach production undetected when the team validating the output lacks sufficient domain knowledge. A policy or billing system built on code that looks right but encodes the wrong business rules can create operational and regulatory risk. Getting the architecture or business logic wrong is expensive to reverse. An internal build reflects the knowledge, assumptions, and experience of the team making those decisions at that point in time. A vendor platform shaped across hundreds of implementations, by contrast, can incorporate patterns tested at scale, edge cases encountered in practice, and business logic refined across lines of business and regulatory environments. The Real Build vs. Buy Calculation in 2026 The build-vs-buy question for insurance core systems hasn’t disappeared. For some use cases, the answer has shifted. For point solutions, internal tools, and workflow automation, AI has made building faster and more economical. The calculation changes when it moves to the system of record. AI amplifies what’s already there. The quality of the data, the depth of the domain knowledge, and the rigor of the validation process all shape the outcome. A carrier with strong internal expertise in a specific area can build powerful tools on top of a capable platform. Building and maintaining the platform itself compounds a different kind of cost. The carriers best positioned for the next decade will be deliberate about where they build versus where they buy, using internal resources where technology can create meaningful differentiation and relying on proven platforms where complexity has already been solved. Origami Risk gives carriers a foundation built on that logic: a platform developed across hundreds of implementations, paired with the flexibility to support the workflows, tools, and capabilities that are specific to each carrier. The expertise needed to validate the system, surface edge cases, and keep pace with change doesn’t come from a code generator. It comes from a partner who has already built for it. Download the carrier build-vs-buy guide to see how leading carriers are thinking through the decision.