Agentic AI for insurance claims modernization on AWS

Modernising the insurance claim management system with HCLTech Agentic AI mainframe solution

5 min 所要時間

HCLTech leveraged multiple AWS services to accelerate the migration program for a major UK-based general insurance company, primarily focused on personal lines such as motor, home and pet insurance, as well as rescue services and commercial insurance for small businesses

The Challenge

As part of a large-scale claims modernization and mainframe exit program, the insurer faced multiple challenges while migrating legacy claims and policy data to Guidewire ClaimCenter on AWS:

Highly complex legacy data landscape, with decades-old mainframe schemas, semi-structured data, free-text notes and batch-driven processes
Heavy dependency on mainframe SMEs for data understanding, mapping and validation
Manual and time-consuming data analysis and transformation, limiting scalability and repeatability
Difficulty in maintaining consistency and reconciliation across large data volumes
Traditional migration approaches failing to scale, resulting in extended timelines and increased program risk

These challenges highlighted the need for an intelligent, automated and repeatable approach to legacy data understanding and migration.

The Objective

The primary objective was to accelerate the claims modernization program by leveraging Generative AI to automate legacy data analysis, transformation and validation. The insurer aimed to:

Reduce migration timelines and effort
Minimize reliance on scarce mainframe SMEs
Improve data quality, consistency and reconciliation
Enable a scalable and repeatable migration factory
Enhance productivity across multiple SDLC personas

The Solution

HCLTech designed and implemented an Agentic AI–based mainframe modernization accelerator built entirely on AWS-native services, leveraging Amazon Bedrock’s multi-agent framework to automate legacy data understanding, transformation and validation for the Guidewire ClaimCenter migration.

The solution was architected to enhance productivity across the full SDLC by enabling collaboration between specialized AI agents and human personas through a secure, scalable cloud platform.

AWS-powered Agentic AI platform

Amazon Bedrock (Multi-Agent framework): Used to build and orchestrate multiple AI agents, each responsible for a specific migration task, while enforcing Responsible AI guardrails for toxicity control, PII redaction and hallucination mitigation
Orchestrator Agent (Supervisor): Coordinates user requests and routes tasks across collaborating agents using Amazon Bedrock Agents
Transformation Agent: Analyzes legacy mainframe data stored in Amazon S3, generates source-to-target mappings, transformation logic, ETL scripts and supports code transformation from Cobol and JCL to modern languages such as Java
Database Agent: Generates SQL queries and validates transformed data within Amazon RDS, ensuring data parity between legacy systems and Guidewire ClaimCenter

User interaction and integration layer

Interactive React UI hosted on ECS Fargate: Provides a conversational interface for analysts, developers, QA teams and business users to interact with AI agents
Amazon API Gateway (WebSocket API): Enables persistent, real-time communication between users and the Bedrock agents
AWS Lambda: Processes user requests received via WebSocket APIs and invokes the appropriate Bedrock agents dynamically

Data, security and governance

Amazon S3: Stores legacy data extracts, transformation artifacts and intermediate outputs securely
Amazon RDS: Hosts transformed relational data used for validation, reconciliation and UAT review
AWS identity and access management (IAM): Enforces fine-grained, least-privilege access across users, agents and AWS services
AWS Key Management Service (KMS): Provides encryption at rest for data stored in S3 and RDS
AWS Control Tower: Ensures centralized governance, compliance and guardrails across multi-account AWS environments
Amazon CloudWatch: Enables monitoring, logging and operational visibility across agents, Lambda functions and APIs

This AWS-native, Agentic AI–driven solution transformed a traditionally manual and SME-heavy migration effort into a scalable, repeatable and secure migration factory, significantly accelerating the insurer’s claims modernization journey.

AWS Services Used

Amazon Bedrock Agents
Amazon API Gateway
AWS Lambda
Guidewire Cloud Platform (GWCP) hosted and operated on AWS
Amazon CloudWatch
AWS Key Management Service (AWS KMS)
Amazon Simple Storage Service (Amazon S3)
Amazon Relational Database Service (Amazon RDS)
AWS Identity and Access Management (AWS IAM)

The Impact

Compared to the traditional migration approaches for large-scale mainframe modernization, the Agentic AI solution accelerator reduced the effort and timelines by 30%. In specific use cases, data mapping has achieved benefits of up to 90%.