From Large Language Models to Reasoning and Agents
Journal article, 2026
Software engineering is shifting from standalone large language models to reasoning-driven and agent-based systems. Drawing on ASE 2025 research, this column highlights how AI agents now test, debug, adapt, and explain software, making a transition toward more autonomous and observable AI-assisted development.
Testing
Cognition
Computer bugs
Software
Software engineering
Debugging
Modeling
Codes
Cognitive systems
Large language models
Author
Miroslaw Staron
University of Gothenburg
Chalmers, Computer Science and Engineering (Chalmers), Software Engineering (Chalmers)
Silvia Abrahao
Polytechnic University of Valencia (UPV)
IEEE Software
0740-7459 (ISSN) 19374194 (eISSN)
Vol. 43 3 92-96Subject Categories (SSIF 2025)
Software Engineering
Computer Sciences
Computer Systems
DOI
10.1109/MS.2026.3662034