Why agentic AI will transform insurance claims assessment

A new tool can assess claims against rules, compare outcomes against past cases and ensure decisions are auditable and transparent.

In brief:

• Australians are wary of AI making claims decisions, but skills shortages and escalating costs mean insurers need to harness this game-changing technology.
• Trust issues can be resolved by Agentic AI documenting initial decision reasoning and generating audit trails to meet regulatory and customer expectations.
• By building policy terms, rules and claims philosophies into an agentic AI workstream, insurers can achieve both return on investment and improve the customer experience within 12 months.

According to the EY AI Sentiment Index Study, Australians have substantially lower comfort with AI than many other consumer cohorts around the world. And nowhere is this more apparent than when it comes to evaluating insurance claims. The Study found that less than a third (31%) of Australians are comfortable with insurers using AI in this manner – compared with the 48% global average. The big issue is that 58% of Australians say it’s a major concern when organisations don’t explain how AI is making decisions.

This sounds like bad news for insurance companies hoping that AI will resolve the issues being caused by increasing skills shortages in their claims assessment teams. As claim volumes grow in both size and complexity and experienced assessors retire, insurers are under mounting pressure to maintain accuracy, fairness, compliance and turnaround times.

The irony is that AI is uniquely suited to ease pain points in what is now an insurer’s most labour- and time-intensive function. Claims assessors must review and interpret hundreds of pages of complex documentation, including medical reports, case histories and the fine print of historical policy terms and conditions. Every decision must be fully traceable, with clear audit trails to meet both internal governance standards and external regulatory requirements.

Agentic AI can rapidly capture and synthesise lever-arch files of structured and unstructured information. Incorporated within claims workflows, this technology can seamlessly support and accelerate the claims process. Yet, until recently, the question of trust – and the mammoth task involved in building bespoke AI tools – have prevented insurers from taking up what could be a game-changer.

A trusted asset for modern claims management

The EY claims assessment tool has been purpose-built by a team of life insurance claims professionals to directly address the real-world challenges faced by claims teams managing individual and group claims across TPD, IP and Trauma. The tool offers insurers the pre-built capability to analyse and summarise claims information, with logic and data structures, guardrails, and compliance with local regulations. Insurers then work with EY teams to make the model their own, ingesting product and policy information, historical claims outcomes, while overlaying their claims philosophy, rules and processes.

Now the tool is operational in the Australian market, local claims teams doing 10-week proof of value tests are discovering a step-change in speed, consistency and insight.

Rapid claims summaries and preliminary validations

Claims teams are finding that the tool dramatically accelerates the review of complex claims by extracting and evaluating relevant information from diverse sources, including policy documents, claim forms, medical reports, call transcripts or audio files, and databases. Claims assessors can query the summaries the tool provides – asking for specific diagnosis details or checking historical policy wording – without having to comb through hundreds of pages of otherwise unsearchable documents. The tool cross-references information across sources to identify inconsistencies or missing data, flagging areas that may require further investigation. With policy terms and exclusions integrated into the logic, it can perform initial validations and claims decisioning, including triaging complex claims.

All the while, the tool automatically documents its reasoning, backed with reference sources, generating audit trails that comply with current regulations.

Moreover, the claimants’ experiences are improved, with streamlined support through the initial information gathering stage and material reductions in claim decisioning and payment times.

Maintaining consistency with historical cases

Insurers can find it challenging to maintain alignment with claims philosophy and historical outcomes, especially as teams scale or evolve. The tool helps to provide confidence in consistency by comparing new claims against historical cases with similar facts and circumstances. This enables claims assessors and quality assurance teams to identify precedent, apply consistent reasoning and document decisions with greater clarity, which in turn, helps to minimise claims leakage.

The tool also learns from every interaction with an assessor, enabling insurers to capture a generation of experience and knowledge before the current cohort of claims assessors begins to retire.

Keeping up with regulatory and technology changes

When accessed as a service, the tool is automatically updated with changes in legislation, regulation, industry codes (LICOP) and internal policy. Claims teams can remain compliant without changes disrupting workflows.

Also, as LLMs continue to develop, the latest versions will be incorporated in updates of the tool as part of the service.

Allocating human and agentic resources to deliver value

Beyond day-to-day processing, the tool provides senior leaders and quality assurance (QA) teams with operational insights. QA teams get confidence indicators and real-time visibility into emerging trends, risk indicators and decision patterns across the claims portfolio, allowing them to focus on higher-risk claims or identify patterns in decisions made by an individual assessor.

Meanwhile, the tool assesses aggregated data to surface insights into systemic bottlenecks, like claims features that consistently face delays or common points of failure. Armed with this information, managers are able to refine processes, adjust resourcing and focus assessor training on pain points.

The information can also be used in stakeholder communications. For example, supporting and enhancing a policyholder’s claims experience or providing metrics for regulatory reporting.

How to scale without scaling costs

Agentic AI tools like this offer claims teams a way to increase aggregate capability without the cost of additional headcount, supplementing attrition and supporting growth ambitions. The tool enables less experienced handlers to manage many more cases, supported by an intelligent assistant, and frees up experienced claims assessors to work on complex cases.

This is how insurers can pursue growth with confidence, knowing their claims operations can keep pace.