Modern software delivery pipelines face growing complexity and manual overhead when authoring code scaffolds, tests, and infrastructure configurations. We introduce “DevGen” an integrated Generative-AI(GenAI) assistant that embeds large language models at four pivotal stages of a CI/CD workflow, which includes feature breakdown, code templating, automated test synthesis, and pipeline authoring. By integrating DevGen into a GitHub Actions environment, we compare its performance against a traditional toolchain over multiple sprints. Our empirical analysis demonstrates a 30 % reduction in story completion lead time, a 25 % increase in automated test coverage, and measurable improvements in developer satisfaction. We conclude with recommendations for domain-specialized fine-tuning, closed-loop feedback, and security policy generation to further enhance AI-driven delivery.
Amar Gurajapu (Wed,) studied this question.