Project Duration: 2026
Budget: SEK 359,000
Partners: Grundfos, Volvo Cars, Zenseact, Axis Communications
Participating researchers: Vasili Mankevich, Aleksandre Asatiani
The aim of this project is to understand how GenAI augments software engineering work outside of coding. The project will work on creating a new evidence-based metric foundation for AI-era engineering.This project examines two urgent questions for software organizations. First, we investigate whether widely adopted metrics (e.g. DORA) remain meaningful when GenAI accelerates development or alters workflows. Second, while most attention is on AI-assisted coding, much of software development—particularly in embedded contexts—happens in non-coding tasks such as documentation, requirements work, testing coordination, debugging, and communication. These tasks are essential but rarely measured. The project maps how developers actually spend their time, identifies where GenAI helps or could help, and develops metrics that reflect this broader, often overlooked part of development work.
To this end we will:
- Evaluate whether widely used metrics such as DORA and cycle time remain meaningful in GenAI-augmented development.
- Map how embedded software teams actually spend their time, what are the value- adding activities, and investigate how GenAI shifts or reduces non-coding work.
- Develop GenAI-ready metrics that capture the emergent practive of hybrid human–AI development.