Generative AI is making it faster and cheaper for insurers to develop their own technology, but Earnix warns that lower development barriers do not remove the long-term costs of governance, maintenance and regulatory accountability.
The InsurTech firm argues that insurers reassessing their technology strategies risk focusing too heavily on how quickly they can create a working solution, rather than what it takes to operate it reliably at scale. AI can generate code, accelerate model development and turn prototypes into applications in a fraction of the time, prompting insurers to reconsider whether buying established software still makes sense. However, Earnix says creating a model or calculating a rate is only part of the challenge. Insurers must also turn these capabilities into reliable, governed and scalable processes that support business decisions over the long term.
AI-assisted development has made it easier for insurers with strong actuarial, data science and technology teams to build their own software. Yet once an application enters a production environment, the organisation becomes responsible for its ongoing operation, including testing, security, governance, auditability, approval workflows, performance, regulatory requirements, maintenance and further development. These responsibilities continue long after launch, meaning that comparing the upfront cost of an internal build with the price of an external product provides an incomplete picture. Insurers must also consider the resources required to maintain, govern and improve internally developed systems over several years.
This also raises questions about how insurers use their specialist teams. Actuaries create value by assessing risk, evaluating assumptions, developing pricing strategies and responding to changing market conditions, while technology teams balance competing demands across the wider business. When these specialists are also responsible for maintaining production software, conducting quality assurance, updating governance controls and responding to new requirements, their attention can shift from strategic priorities towards keeping systems operational. As internal priorities change, developers move roles and budgets are redirected, systems that received significant attention during development may become harder to maintain. Earnix argues that the decision is therefore not simply whether an insurer has the skills to build a solution, but whether doing so represents the best use of its resources.
The growing role of AI in software development has also raised questions about whether the technology can validate the systems it helps create. While AI is expected to play a greater role in development and testing, insurers must still demonstrate how models and processes were developed, which changes were approved, what testing was completed and how decisions can be explained or reconstructed. AI-generated code and automated testing do not remove these obligations, nor does the choice between purchasing software, developing it internally or using AI to create it change an insurer’s responsibility for the systems it deploys. As AI becomes more embedded in insurance workflows, Earnix argues that governance becomes increasingly important, with firms needing to show that systems continue to perform as intended as models, regulations, market conditions and business requirements evolve.
Insurers have traditionally turned to in-house development to accommodate their individual products, risk appetites, pricing strategies, workflows and competitive priorities. However, Earnix argues that bespoke requirements do not necessarily mean every layer of technology must be built from scratch. Established insurance technology platforms can provide infrastructure for modelling, pricing, rating, deployment and governance while allowing insurers to apply their own data, models, strategies and business logic. Rather than choosing between developing an entire capability internally or purchasing a complete external solution, insurers can identify where differentiation matters most and where proven infrastructure can provide a reliable foundation.
Earnix says insurers should take the opportunities presented by AI seriously while ensuring technology decisions account for more than development speed and upfront cost. Key considerations include ongoing maintenance expenses, who will own a system in three to five years, how it will adapt to regulatory changes, how updates will be tested and governed, and whether it can scale as demand increases. Insurers must also consider how much internal capacity will be diverted from advancing business strategy to maintaining technology.
Earnix argues that while the technology needed to build software is becoming more accessible, the discipline required to operate insurance decisioning safely, consistently and at scale remains considerably more difficult to replicate. For insurers reassessing their technology strategies, the key question is not simply whether AI makes building easier, but whether an internally developed solution can be maintained, governed and improved effectively throughout its lifecycle.
Copyright © 2026 FinTech Global
