Implementation of DevSecOps practices with artificial intelligence to strengthen security and efficiency in software development in a software factory company

Abstract

The growing imperative to integrate security at the earliest stages of software development has led to the adoption of DevSecOps practices augmented by artificial intelligence (AI). The study assessed the impact of implementing these practices on the security and efficiency of software development within a company in Lima during 2025. A continuous integration and continuous delivery (CI/CD) pipeline was designed and implemented with automated security and quality controls. The pipeline incorporated quality gates via SonarQube, dependency scanning with Snyk, and AI assistance through GitHub Copilot, following the PHVA methodology. A pre-experimental design with pre- and post-measurements was used with a selective sample of 25 deployments. The study evaluated the density of detected findings (DF), detection time (DT), and correction time (CT). The results indicated a 198,7% increase in the density of detected findings, a 99,1% reduction in detection time, and an 86,4% decrease in correction time. All differences were statistically significant for all three indicators (p< 0,001; r ≈ 0,87). Integrating DevSecOps practices with AI strengthens security, improves development efficiency, and provides a replicable implementation strategy to advance the maturity of secure development processes.