Thought Machine and AWS automate legacy core banking conversions

Thought Machine, the London-based core banking platform, has partnered with Amazon Web Services to automate the conversion of legacy mainframe code into cloud-native financial products, announcing the initiative on 29 September. The partnership combines AWS Transform (the hyperscaler's reverse-engineering tooling for legacy systems) with Thought Machine's Vault Forge, an AI suite powered by Amazon's Bedrock language models.

The collaboration addresses a constraint that affects every major European bank: millions of lines of COBOL code embedded in decades-old mainframe systems, typically handling account management, payment processing and risk calculations. Migrating these systems has historically required years of manual translation, line-by-line conversion and extensive testing. Thought Machine said the new pipeline compresses this process into days by extracting business rules as structured specifications and using AI agents to synthesise simulation-tested Python code directly deployable on its Vault platform.

The four-stage pipeline operates entirely within a bank's secure AWS environment. First, the extraction phase identifies business logic in legacy applications and outputs structured technical specifications. Next, consolidation merges hundreds of redundant product variations (the legacy of decades of ad-hoc patches) into lean, modern business rules. Vault Forge then orchestrates AI agents to translate these specifications into SDK-compliant Python financial products, including test suites and platform configurations. A human-in-the-loop validation stage gives bank engineers authority to review and sign off each converted product before it moves to a Vault sandbox for further testing.

Paul Taylor, Thought Machine's chief executive and co-founder, said the partnership "fundamentally rewrites the economics of core modernisation." He added that banks can now deploy AI to convert legacy specifications into cloud-native products with the speed and audit rigour needed to eliminate technical debt.

Charith Mendis, AWS's head of banking and financial services, said the solution addresses what banks see as their hardest modernisation challenge. "We believe it will fundamentally change how the industry approaches core transformation," he added.

Thought Machine is simultaneously launching a Core Modernisation Accelerator programme with AWS, pairing technical architects, product specialists and its own forward-deployed engineers directly with bank teams to oversee delivery, governance and architecture design during migrations.

The timing reflects Thought Machine's momentum in Europe. In August, Deutsche Bank's Private Bank selected Vault Core under a 10-year agreement to replace 15 legacy systems with two unified platforms, targeting €300 million in annual savings by 2028. The same month, Pathward, the U.S. sponsor bank, also selected Vault Core to modernise its own lending and payment infrastructure. Thought Machine now counts over 65 leading banks worldwide as customers, including several strategic investors. The company has raised more than £500 million to date and operates from offices in London, New York, Singapore, Sydney, and Lisbon with a team of 497 people.

By combining its existing Vault platform with AWS's modernisation tooling and frontier AI models, Thought Machine shifts its pitch beyond replacing just the core systems with new business need. The partnership enables banks to migrate their entire legacy infrastructure—not just add new services on top of it—with machine-verified correctness and a fraction of the manual effort. For institutions holding petabytes of COBOL logic, that changes the calculus on modernisation from optional innovation to feasible survival.