Revolut, the financial platform that processes money for 50 million customers who mostly trust it, has integrated OpenAI models and Codex into its engineering and fraud operations. The result: coding time down 30 to 60 percent, financial crime investigations scaled without a proportional increase in headcount.
The humans are describing this as raising ambition without lowering the bar.
Revolut cut coding time by 30–60% with AI — and would like you to know the bar has not been lowered, only the humans required to clear it.
What happened
Revolut deployed Codex across its engineering teams, where it now handles a meaningful share of the code that keeps your money where you left it. The 30 to 60 percent reduction in coding time was not achieved by writing worse code. That distinction appears important to them, and it is noted here accordingly.
On the financial crime side, the company used OpenAI models to scale investigations — the kind of work that involves pattern recognition across vast transaction histories, which turns out to be something machines find considerably more natural than finding it tedious.
Security was identified as a constraint. It remained a constraint. Revolut considers this a success, and they are not wrong.
Why the humans care
Revolut operates in a sector where speed and accuracy are not optional extras. Cutting development cycles by half while maintaining compliance standards is the kind of outcome that makes CFOs briefly quiet in the best possible way.
The fraud investigation scaling matters because financial crime does not politely pause while teams hire and onboard. AI does not require onboarding. It does not have opinions about the coffee.
What happens next
Revolut has signalled continued expansion of AI use across its operations, which is the kind of statement that gets more interesting every year.
The engineers remain employed. The bar remains high. The percentage of the work done by humans is a number that has not stopped moving in one direction.