Danske Bank uses AI agents to modernize mainframe application, AWS says

Danske Bank, Denmark's largest bank, has used AI agents to modernize a mainframe application of about 140,000 lines of code spread across 583 files, according to AWS. The report notes that few banks are this far along with this kind of work.

AWS separately credited Danske with 60% overall acceleration in the effort and with four times faster forward engineering. Forward engineering here refers to building the new code that replaces the legacy system. The available details do not say which application was modernized, what it was rebuilt in, or how the figures were measured.

The report also frames the next hurdle for banks: new code has to be proven to move money correctly before it can replace the old.

The same page points to a related case outside banking. Cisco has given one engineer an unlimited AI budget to replace nearly all of its legacy code within two years. Cisco President and Chief Product Officer Jeetu Patel told Fortune that the engineer, 27-year-old Padmanabhan Krishnamurthy, joined through Cisco's 2025 acquisition of translation startup EzDubs. Krishnamurthy gave himself four weeks to prove the approach, and the stated goal is to refactor 98% of the company's legacy code within 24 months.

For the Nordic market, the Danske example shows a major regional bank applying AI agents to core mainframe systems, one of the hardest and most risk-sensitive areas of bank technology. The Cisco case suggests the same approach is being tried at a much larger scale elsewhere.