On 16 November 2026, AWS announced a systematic, AI‑assisted methodology that transforms legacy mainframe Assembler applications into modern Java Spring Boot applications on AWS cloud infrastructure.
The approach, designed to address the high maintenance costs and staffing shortages that plague Assembler systems, uses Amazon Bedrock and Kiro IDE to extract business rules from Assembler code. Subject‑matter experts validate these rules before the process moves to the next stage.
Five‑Phase Transformation Workflow
The workflow comprises five distinct phases: (1) rule extraction with AI, (2) creation of detailed functional specifications, (3) conversion of Assembler to COBOL as an intermediate step, (4) functional equivalence testing of the COBOL code, and (5) automatic conversion of validated COBOL to Java Spring Boot by AWS Transform.
Each phase incorporates human‑in‑the‑loop validation checkpoints to capture domain‑specific logic and edge cases, ensuring business logic continuity and reducing rework.
Risk Reduction Through Intermediate COBOL
Using COBOL as a bridge between Assembler and Java preserves the original business logic and provides a familiar language for experts to review. This intermediate step lowers the risk associated with a direct Assembler‑to‑Java rewrite.
Functional equivalence testing is performed incrementally throughout the process, not only at the end, allowing early detection of issues and faster remediation.
The final Java Spring Boot applications are deployed on AWS infrastructure, offering cloud scalability and resilience for banking, insurance and airline customers that historically rely on IBM z/OS, JCL and VSAM data stores.
