EXPLORE PUBLICATIONS BY COUNTRIES

Agentic AI, Cloud-Native Development, Software Engineering, Productivity Improvement, Autonomous Systems.

This research paper identifies the disruptive effect of Agentic Artificial Intelligence (AI) on cloudnative software development and design. As software systems become increasingly complex, conventional approaches to development are strained to the breaking point for velocity, efficiency, and scalability. The advent of goal-directed, self-organizing agents called Agentic AI offers a new paradigm for dealing with these demands. This study investigates if and how a multi-agent system can be utilized to automate and optimize some aspects of the cloud-native development life cycle such as code generation and testing, deployment, and monitoring. Overall, the objective is to quantify productivity benefits and benefits in software quality that are a consequence of incorporating Agentic AI.To complete this research, we employed a synthetic dataset that was generated to model the life cycles of 50 independent cloud-native microservices projects over one year. The data include measurements such as lines of code, bug density, deployment rate, and developer-hours. The essence of our research was deploying a custom-made Agentic AI framework in Python coded and developed using LangChain on top of major industry cloud platforms such as Kubernetes and AWS. It enables specialized agents for code generation, task decomposition, security scanning, and performance testing. When we compare project development metrics of projects that have been developed using this approach and a control group of projects that use traditional CI/CD practices, we demonstrate a dramatic reduction in the development time and an astronomical growth in the quality of the code. Results drawn in this research provide emphatic proof of the efficacy of Agentic AI in terms of cloud-native app development as a scalable and viable solution compared to conventional methods. Python data science libraries (Scikit-learn, Pandas) and visualization libraries are used as tools for analysis to render output.