AI-Powered Software Development: Transforming the SDLC through Intelligent Automation

Abstract

A paradigm shift in software engineering is emerging as artificial intelligence (AI) tools and methods become increasingly embedded within the Software Development Life Cycle (SDLC). This paper introduces AISA-SDLC (AI-Integrated Software Automation - Software Development Lifecycle), a modular framework that integrates intelligent automation, human oversight, governance, and continuous feedback into each phase of modern software development. To ground the framework in real-world practice, a developer-centric survey was conducted to analyse current AI adoption trends, usage patterns, and trust levels within the developer community. A layer-by-layer prototype was created and evaluated based on these observations using easily accessible open-source techniques, such as large language model (LLM) code generation, static linting, vector similarity search using FAISS and Sentence-BERT embeddings, lightweight runtime observability using Prometheus and Grafana, and natural language prompt structuring. Even while AI greatly speeds up typical coding jobs, explainability, governance checks, and layered human oversight are still necessary to guarantee trust, security, and maintainability, according to a comparison with a traditional workflow. The study concludes by outlining current limitations and future directions for scaling AISA-SDLC into a fully integrated, responsible AI-augmented development pipeline.