O papel da inteligência artificial no apoio ao desenvolvimento ágil de software: uma revisão da literatura

Abstract

Context: The use of Artificial Intelligence (AI) has expanded significantly within the field of soft- ware development, with applications ranging from the automation of repetitive tasks to decision- making support. At the same time, agile methodologies have become predominant in develop- ment teams, promoting iterative cycles and continuous delivery. The combination of these two elements has drawn the attention of both the scientific community and the industry, especi- ally due to their potential to optimize processes, reduce rework, and increase team efficiency. However, there is still a lack of consolidated studies that systematically reveal how Al has been effectively used to support agile software development. Objective: This project aims to identify and analyze the main ways in which Al has been applied to support the agile software develop- ment lifecycle. Method: To achieve this, a Systematic Literature Review (SLR) was conducted based on studies published in scientific databases, using rigorous inclusion and exclusion cri- teria. Results: The analyzed studies show a predominance of machine learning and Natural Language Processing (NLP) applications, primarily supporting tasks such as estimation, predic- tion, and planning. The main benefits include increased estimation accuracy, task automation, and decision-making support. The most recurrent challenges involve data quality and the limited generalization capacity of models. The literature also highlights a focus on Scrum and general agile contexts, as well as growing interest in Generative Al and Large Language Models (LLM).