Articles

As insurers struggle with fragmented data and slow analytics cycles, AIA's PRISM project shows that the solution may already sit within existing technology stacks

AIA Shared Services, the shared services arm of pan-Asian life insurer AIA Group, was honoured with the Technology Integration Excellence Award at the Insurtech Connect Asia Awards on 1 July 2026.

However, the winning solution, Project PRISM, was not built from scratch by an internal engineering team. Nor did it depend on purpose-built AI models trained on insurance-specific data. Instead, Project PRISM runs on Microsoft Fabric, Azure OpenAI and the Power Platform suite, tools available to any enterprise with the right licensing agreement.

This has made Project PRISM cost-effective, allowing AIA to implement the solution using low-code and no-code capabilities, reducing reliance on proprietary tools and their maintenance, while simplifying integration within a unified technology ecosystem.

“This approach enables faster time-to-value while maintaining cost discipline, strong governance and alignment with enterprise technology standards,” said Gordon Ew, Head of Policy Servicing at AIA Shared Services.

PRISM is built on Microsoft Fabric's Lakehouse and OneLake architecture, which consolidates structured and unstructured enterprise data into a single governed environment.

The solution aims to resolve a commonly cited operational constraint in the insurance sector, where large multi-market carriers juggle fragmented analytics landscapes characterised by siloed datasets, manual data collection and static reporting cycles.

Data that should inform underwriting, claims, customer servicing and portfolio management often sat in disconnected systems and was only accessible through labour-intensive extraction and reconciliation processes, slowing teams down.

To move away from that model, data pipelines now run through automated notebook execution, reducing the manual handling that previously characterised analytical workflows and improving the repeatability of outputs.

The system maintains what is described as a “single source of truth”, a governed foundation that ensures teams across business units are working from consistent, validated data rather than divergent local copies.

PRISM also adds generative AI capabilities through Azure OpenAI, which handles tasks including data classification, mapping, formatting and validation workflows, automating processes that would otherwise require analyst time and introduce human error.

Outputs are delivered through Power BI dashboards, Power Apps interfaces and agent-assisted workflows built in Microsoft’s Copilot Studio, embedding insights directly into the operational tools teams already use.

“Together, these improvements enable teams to focus on higher-value activities, while ensuring faster delivery of insights and more consistent output quality,” Gordon said.

The insurer also ensured that the connection between Microsoft Fabric and Azure OpenAI was handled through secure configuration patterns and aligned with internal security standards that prohibit fully public AI models for enterprise use cases, mandating controls when regulated data is involved.

Off-the-shelf scaling

Building Project PRISM on a unified Microsoft Fabric architecture means the platform is highly scalable, Gordon added.

Project PRISM was designed to expand across business units and markets without requiring infrastructure overhauls and can absorb new datasets and AI use cases as operational demands grow. It is also engineered to handle increasing data volumes and transaction complexity without performance degradation.

The right architectural decisions, combined with organisational discipline, mean insurers can use off-the-shelf AI models to build solutions that scale efficiently.

“With the right governance in place, the platform can be scaled in a consistent and controlled manner,” said Gordon. “We also continue to expand its use across more end-to-end processes to maximise its long-term value.”

The implementation of automated data pipelines and AI-enabled workflows has delivered operational efficiency gains across AIA Shared Services, Gordon added. Manual analysis and repetitive processing tasks have been reduced through automation, while reusable data models and structured pipelines have streamlined broader operations.

Turnaround times have improved, and the effort required for data validation, cleansing and reconciliation, processes that consume significant analyst time in most large insurance organisations, has been reduced.

“Overall, the shift to a more integrated and data-driven approach has helped improve both efficiency and service quality, contributing to better customer outcomes over time,” said Gordon.

Piecing data fragments together

By designing a platform that is market- and business-unit-agnostic, AIA has enabled expansion across its regional operations without replatforming.

With the entire stack built on enterprise-approved technologies that AIA already had access to, the insurer was able to accelerate the adoption of advanced analytics, enabling consistent operational intelligence and future-proofing the organisation for growth and innovation.

AIA plans to expand the platform to cover end-to-end case flows, add automation triggers and track AI accuracy as the system scales.

“Automation and AI-driven validation help ensure outcomes are consistent and dependable,” said Gordon. “For policyholders, this means shorter turnaround times, fewer errors, and a more reliable experience, especially during important moments in their journey.”