Thought Machine, AWS Tap AI to Help Banks Move Beyond Legacy Systems

The core of this solution lies in its ability to extract business rules from legacy mainframe code rather than attempting a line-by-line translation. By identifying the underlying logic for functions like interest calculations and loan repayments, the process bypasses the complexities of direct code conversion. This approach allows AI agents to focus on converting these abstracted rules into Python-based financial product code, which is then deployed within a bank's secure AWS environment.

The methodology involves AWS Transform analyzing legacy applications to produce structured specifications of business logic. Vault Forge then uses AI models via Amazon Bedrock to generate new financial products, test suites, and platform configurations. This consolidation of redundant legacy product variations before AI generation is intended to streamline the modernization effort.

A key element highlighted is the retention of human oversight, with bank engineering and risk teams maintaining sign-off authority through review stages. This human-in-the-loop approach is crucial for validating AI-generated products before they are deployed to a Vault sandbox for further testing. Thought Machine claims this process can compress multi-year migrations involving millions of lines of code into days, addressing a significant pain point for large financial institutions seeking to shed technical debt.

The joint Core Modernisation Accelerator programme, also launched by the partners, suggests a hands-on support mechanism. It will involve technical architects and engineers from both Thought Machine and AWS working directly with bank teams. This indicates an understanding that the technology alone may not be sufficient without integrated implementation and governance support, particularly for complex transitions.