For insurers, the promise of AI is clear: faster decisions, built on superior insights, leading to improved customer experience. Yet the transition from pilot to enterprise-wide AI adoption continues to be a major pain point.
Decades of complex, interdependent legacy systems slow progress, while intensifying regulatory focus on data, bias, and explainability necessitate strict guardrails.
The real challenge, therefore, is not experimenting with AI, and its subset, generative AI (GenAI), but embedding it within modern, secure, and scalable cloud environments.
Rethinking risk and resilience
Insurers have long used machine learning for risk pricing, but GenAI marks a decisive leap forward.
Foundry research shows that IT decision-makers agree, with the majority believing the technology will improve product development (61%), enable better outcomes (56%), and free up workers to innovate (58%).1
Stephen Holdstock, Chief Technology Officer of EMEA Insurance at EPAM Systems says: “We’re at the very start of the GenAI journey, but already it’s helping with time-consuming tasks such as researching engineering reports and benchmarking.”
It is also enabling underwriters to instantly summarize submissions, interrogate vast data sets, and determine whether risks fit underwriting criteria.
Holdstock explains there are also new AI risks, however, with threat actors using the technology to generate fake research, impersonate individuals, and leak data.
Building customer trust
GenAI can also build trust by demystifying complex policy documents, making obligations clearer, and helping customers test scenarios in plain language.
The need for trust extends to how personal data is used. Wearables in health insurance, for example, offer premium discounts in return for continuous data sharing, but this exchange needs to be carefully managed.
Mik Quinlan, Senior Director at EPAM Systems, says customers want personalization, but they also want control. “Customers should have fine-grained control over their data,” he says, “authorizing access rather than giving it away.”
From legacy systems to cloud-enabled platforms
Modernizing insurance estates is complex. Core platforms are woven into ecosystems spanning ledgers, claims, compliance, and reporting. A simple lift and shift to the cloud is rarely an option.
The pragmatic approach begins with mapping current systems and grouping them by domain (for example, underwriting vs. policy servicing). From there, insurers can create an architecture vision built on API-first design (ensuring clean integration), and the principle of “buy for core, build for experience.” In other words, purchase commoditized central systems, but custom-build those that shape customer journeys.
Incremental migration is best supported by replacing old functions step by step, rather than through a risky big-bang cutover. As Quinlan observes: “If the primary motivation is to innovate, target the systems that block that innovation first — even if full modernization takes years.”
Responsible AI in a regulated market
Regulators are sharpening their focus on AI which means governance frameworks must keep pace.
“If an AI system is making a recommendation, it must be able to explain why, the same way an underwriter records their rationale,” says Holdstock.
Quinlan adds: “Start with your enterprise risk and governance policies, update them for AI, and walk through the process before any system goes live. This ensures all relevant stakeholders are fully informed how the system was built and tested, including any measures taken to adhere to compliance directives.”
Embedding human oversight, bias testing, and version-controlled policies ensures that AI adoption enhances resilience, safeguards compliance, and protects business reputation.
Conclusion: Modernization as the path to AI success
AI is no longer a side experiment for insurers, it is the next stage of modernization. Success lies in embedding AI within cloud-enabled platforms that are secure, scalable, and resilient.
EPAM and Microsoft Azure bring together the tools to assess legacy estates, prioritize migration, and modernize responsibly. Together they enable insurers to adopt AI at scale, strengthen resilience, and deliver better outcomes for customers.
1 Foundry Cloud Computing Study 2025