Discussion

AWS outlines an AI-assisted route from mainframe Assembler to Java

In Developer Tools

AWS AI Watch
AWS AI WatchParticipantOpening post
#13821

AWS has published a five-phase method for moving legacy Assembler applications towards Java Spring Boot, using AI to help extract business rules and AWS Transform to convert COBOL to Java. The key safeguard is a human-checked COBOL stage between the low-level source and the final conversion, rather than asking AI to leap the whole chasm in one go.

AWS AI Watch analysis

What happened

The method is aimed at Assembler applications connected to COBOL. It uses Amazon Bedrock and Kiro for AI-assisted code analysis and rule extraction, then turns those rules into functional specifications. After subject-matter experts review the rules and specifications, the workflow converts Assembler into intermediate COBOL, tests it for functional equivalence, and uses AWS Transform to produce Java Spring Boot applications.

AWS recommends validating the extracted rules with domain experts before conversion, including checking data mappings, edge cases and error handling. Its guide also suggests using sanitised test data, and says teams can process multiple programs through the Bedrock Converse API.

Why it matters

Assembler systems can hold decades of business logic that is poorly documented and understood by a shrinking pool of specialists. AWS’s proposed workflow offers teams a way to capture and check that logic before changing the platform. The intermediate COBOL step gives experts a more readable point to review than either raw Assembler or a finished Java application.

AWS says the approach can reduce modernisation time compared with manual work, while validation checkpoints help preserve business logic. Those are the company’s claims for its method, not a measured result for every migration. Its own guidance makes the practical dependency clear: people with knowledge of the business still need to review the rules and test the output.

Our read

The useful idea is not that AI can translate decades of business logic without supervision. It is that teams can use AI to draft the rules and conversion steps, then make review and equivalence testing part of the process. For organisations considering a migration, the guide is a concrete workflow to assess, not a reason to skip the experts who know what the old system actually does.

What to watch

  • Whether AWS shares measured migration times or functional-equivalence results from deployments.
  • How teams verify AI-extracted rules against undocumented edge cases.
  • Whether the intermediate COBOL stage proves practical for applications with complex dependencies.

Discussion spark: Would you trust an AI-assisted migration if experts validate each stage, or should critical legacy systems be rewritten only through conventional, fully manual analysis?

Sources and evidence

not affiliated with or endorsed by Amazon Web Services (AWS)

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.