NNIT shines a light on decades of public sector IT systems with a new AI method

NNIT launches the AI method, Lumina, to help public organisations modernize complex and ageing IT systems with lower risk and greater efficiency.

Public digital solutions support a large part of everyday life in society – from case handling and benefit payments to control, operations, and debt collection. At the same time, many of these systems are complex and difficult to maintain, particularly in light of a growing cyber‑threat landscape and rapid advances in AI, which increase the need for modernization.

A key challenge in modernizing public sector applications is that systems often function as “black boxes”. Documentation is incomplete or outdated, and critical knowledge risks being lost when suppliers or employees change. This can lead to longer development cycles, higher risk, and slower improvements for both citizens and public employees.

Lumina is developed with responsible use of AI in the development and modernization of public applications. The method is designed to ensure traceability, quality, and human oversight throughout the development process, while remaining independent of specific technologies and vendors.

“When modernizing systems that are critical to society, it is essential that AI is used in a practical and responsible way. Improvements must be implemented faster and more safely, without introducing new risks. That is exactly the need Lumina is designed to address,” says Jens Maagøe, Senior Vice President, Region Denmark, NNIT.

Modernizing critical systems with lower risk

Lumina analyses existing code and related documentation to uncover the knowledge embedded in legacy systems. This creates a documented and transparent foundation for further development and modernization.

“Overview, traceability, and control are crucial when delivering improvements to citizens and employees with lower risk. Lumina helps establish a shared and detailed understanding of requirements, code, and documentation for both human developers and AI agents,” Jens Maagøe adds.

The method enables AI agents to translate requirements and specifications into concrete development tasks, while giving developers the necessary context and insight. This supports a more efficient and coherent development process from the definition of requirements to implementation and testing.

A flexible method without technological lock-in

Lumina is based on well-known standard tools and AI agents, applied in a formalized and consistent manner. The method is not tied to a single AI technology, allowing organisations to choose models and assistants that fit their specific needs and technical environments.

“Lumina is deliberately designed with a focus on independence and flexibility. The method is continuously refined based on practical use and technological developments, through an internal governance model inspired by open-source principles,” Jens Maagøe concludes.

Learn more about Lumina here.