Integrating Large Language Models into Agile Software Development: A 2023 Perspective on Productivity and Code Quality

Abstract

This paper explores the application of Large Language Models in Agile software development and their potential impact on software developers' productivity and code quality. Since the maxim of Agile practices is speed, teamwork, and flexibility, the further implementation of AI-based solutions, including LLMs, opens new possibilities in the optimization of these already existing advantages. It will address the challenges of integrating LLMs into established Agile processes, assessing the role these models play in addressing common development issues such as tedious work, slow debugging, and communication problems. LLMs enable automation of code generation, streamline debugging workflows, and support documentation, potentially saving significant time in development and enhancing quality. The most crucial findings indicate that LLMs can lead to higher productivity, characterized by an accelerated coding frequency and an increase in quality through fewer mistakes and greater consistency. The paper concludes with practical recommendations for effectively utilizing LLMs within Agile software development teams.