Insurance core system transformation, a key for business continuity

The impending operational crisis: An evaporating talent pool

The demographic decline of mainframe expertise is a major operational threat facing established insurance firms. For decades, large insurers developed bespoke, on-premises systems tailored to their specific claims and policy management workflows. These self-developed legacy mainframes have functioned as the operational heart of their enterprises. 

However, many specialists who built and maintained these systems are reaching retirement age, creating a critical exposure in insurance continuity. We have seen first-hand how unexpected departures from this critical talent pool leave major insurance firms scrambling to maintain their core operational systems. This organizational fragility emphasizes how core system transformations must be reframed from a discretionary technology upgrade to a board-level priority for operational continuity.

Indeed, early quantitative findings from BearingPoint’s 2027 P&C Core Insurance Systems Study³ confirm that this resource constraint has reached an acute stage. The hundreds of European insurance executives surveyed rated the “Availability of IT Resources to maintain and modernize platforms” among their most pressing challenges, scoring it at 7.4 out of 10. This represents a significant increase from the score of 6.6 compared to the 2025 study. With the personnel required to keep these legacy architectures functioning disappearing, insurers face a narrowing window to migrate their operations to standard, maintainable platforms.

No option to leapfrog from mainframe to AI

Our clients consistently ask whether it is possible to bypass core legacy modernization and leapfrog directly to an AI-driven operating model – avoiding the cost and potential disruption of first transforming from a mainframe to a SaaS setup. For the insurance industry in particular, the answer is unequivocally no. The appeal to executives eager to demonstrate rapid innovation is obvious, but the strategy is flawed on both regulatory and technical grounds.

Recent insurance regulation limits AI use by requiring insurers to keep decisions transparent, documented, and accountable⁴. However, many advanced AI models are probabilistic and can be difficult to fully explain, validate, and evidence in regulated workflows without strong controls and a stable operational foundation. That foundation is what allows every AI-driven decision to be traced back to a verified, auditable source of data, rather than a black-box process. For that reason, a standard system of record should remain the operational backbone, with AI layered on top to augment decisioning.

Furthermore, AI is also only as reliable as the data behind it, and legacy mainframes typically store data in fragmented, non-standard, or isolated databases. Deploying advanced algorithms over this landscape risks inaccurate outputs and compliance violations.

There is no shortcut. Insurers must first migrate from self-developed mainframes to standard, SaaS core systems, establishing the clean, unified data foundation that can subsequently feed AI orchestration layers as the technology matures.

2027 study hints at a complex transformation picture

While our 2027 study is still being finalized, early findings suggest a complex systems transformation picture among insurance companies – and that complexity is instructive. The segment of insurers who report that they “need a new solution but are not ready yet” is down from 24% in 2025 to 14%, while those actively planning a new solution have risen from 21% to 35%. We are seeing that commitment to a defined transformation path is becoming the prevailing posture within the sector.

However, two further figures merit closer attention. The proportion of insurers currently implementing a new solution has declined from 36% to 28%, while the proportion who consider their existing solution stable and requiring no transformation has risen from 19% to 23%. 

This trend is worth watching closely alongside the operational and talent risks described earlier in this piece. Perceived stability does not always equate to structural readiness: a mainframe can run without incident for years, right up until the specialists required to maintain it retire or move on. Insurers in this position may wish to pressure-test whether their core systems would remain structurally ready for AI-driven transformation if that reliance on individual employees disappeared tomorrow.

The methodology guiding transformation has also matured. The proportion of insurers pursuing a completely agile implementation has declined from 83% to 64%, with organizations increasingly adopting a mixed-mode approach that combines milestone-driven governance with agile delivery to avoid over-customization and scope creep. The study suggests that cloud strategy has undergone a similar recalibration: expectations that future core systems will be fully public-cloud deployed have fallen from 60% to 41%, with hybrid and even some modernized, secure on-premises arrangements increasing in popularity. This suggests that insurers are reconciling scalability with regulatory obligation. 

Taken together, the data describes an industry that has largely moved past the question of whether transformation is necessary. For the remaining minority hanging on to legacy systems, the real question is not whether their current systems are working, but how long that will remain true.