A provocative prompt asks a generative AI to design a UK-regulated bank, producing Terraform templates and Java service skeletons. This moves beyond simple API-as-a-service to a reproducible system blueprint, where the entire software stack—ledger, compliance, and infrastructure—is generated from a prompt. The core question is whether regulatory and operational burdens can eventually be encapsulated in code.

The author argues this signals a shift where Gen-AI commoditizes system design. While banking's complexity and regulatory requirements remain immense, the trend is clear: if a platform's value is in workflow logic and schema, AI can generate the platform itself. This challenges many "-as-a-service" models, suggesting their core offering may become a thin abstraction over generatable logic.

Practically, launching a real bank requires more than code; it needs licenses, external partners, and accountability. Yet the exercise proves that the hardest part of building complex institutions is increasingly moving from software creation to system operation and risk management. The future may see more institutions designed as executable, cloud-native templates.