UK-based Thought Machine has launched an AI-powered legacy core migration solution with AWS, combining AWS Transform with its Vault Forge tooling.
The migration solution developed with AWS uses AI to move financial product logic from legacy mainframe systems to its cloud-native Vault platform.
According to the companies, the joint pipeline allows banks to convert millions of lines of mainframe code, such as COBOL, into tested financial products within a live sandbox environment. Rather than translating code line by line, the process extracts business intent and runs inside the bank's own secure cloud environment. Thought Machine states that this can compress processes that traditionally take several years into days.
Why product logic matters in core migrations
In legacy cores, financial products are hardcoded into the underlying database infrastructure, which makes migrations slow and risky. Vault takes a different approach, defining all financial products as code written in Python. Product logic, including interest calculations, fee schedules, repayment workflows, and account lifecycles, is separated from the core infrastructure. This decoupling allows AI agents to bypass database-level complexity and translate legacy rules directly into high-level code.
Core functions based on a four-phased pipeline
The pipeline operates within the bank's AWS environment and is structured in four stages. In the extraction phase, AWS Transform performs discovery and reverse engineering on legacy applications, converting business logic into structured Easy Approach to Requirements Syntax (EARS) specifications. The second stage, called consolidation, reduces hundreds of redundant legacy product variations into a smaller set of modern business specifications.
During synthesis, Vault Forge coordinates specialised AI agents through Amazon Bedrock to generate SDK-compliant Python financial products, along with test suites and platform configurations. Finally, validation introduces human-in-the-loop review gates, giving the bank explicit sign-off authority before the generated products are deployed to a Vault sandbox for further evaluation.
Tasks such as scanning millions of lines of code and mapping complex data dependencies are handled through systematic, AI-guided procedures, and banks receive validated migration proof early in the process.
Implementation support and market context
To support adoption, Thought Machine is rolling out a Core Modernisation Accelerator programme. Under the programme, its technical architects, product specialists, and forward-deployed engineers work alongside bank teams on delivery, governance, and architecture setup.
Charith Mendis, Head of Worldwide Banking, Financial Services Industries, Amazon Web Services, noted that the cost of maintaining legacy infrastructure continues to rise, while customer expectations and regulatory demands are evolving, positioning the pipeline as a response to one of the more difficult challenges in banking modernisation.
The Vault platform is currently used by more than 65 banks and financial institutions worldwide, several of which are also strategic investors in Thought Machine.
