JetBrains Report Reveals AI's Limited Role *Inside* CI/CD Pipelines Despite Widespread Adoption
A recent report from JetBrains, drawing on data from four studies including their 2026 CI/CD Tools Adoption Survey, reveals a nuanced picture of AI's role in continuous integration and continuous delivery. While a significant 48.8% of organizations now utilize AI for at least one CI/CD task, a striking statistic shows that only 8% actually run AI-powered steps directly *inside* a build or test process.
This finding is crucial because it indicates that AI's current impact on CI/CD is largely assistive and external to the core pipeline execution. Practitioners are employing AI for tasks such as code review, writing or fixing tests, and diagnosing build failures, which are valuable but do not fundamentally alter the automated build and deployment stages. This suggests a cautious approach to AI integration, where its capabilities are leveraged for analysis and support rather than direct control or modification of the pipeline's operational flow.
This trend fits within the broader context of evolving DevOps practices, where automation is paramount, but trust in nascent AI capabilities within critical paths remains limited. The industry has seen a rapid adoption of AI in various development stages, from code generation to security analysis. However, the reluctance to embed AI directly into CI/CD execution steps points to concerns around determinism, explainability, and potential for introducing new failure modes. The report also notes a rise in self-hosted CI/CD solutions, with nearly three-quarters of developers self-hosting their primary CI/CD tool, indicating a desire for greater control over infrastructure, data, and security.
For practitioners, this means a few key considerations. First, while AI tools can significantly enhance developer productivity and pipeline observability, relying solely on external AI assistance might leave untapped potential for optimizing build times, predicting failures, and dynamically adapting pipelines. Teams should explore how AI could be safely introduced into testing phases for more intelligent test selection or into deployment for predictive risk assessment. Second, the prevalence of self-hosting suggests that organizations are prioritizing control and security, which might also influence their cautious approach to AI integration. As AI models become more robust and transparent, the industry may see a shift towards more embedded AI within CI/CD, leading to truly intelligent and self-optimizing delivery pipelines. Practitioners should watch for advancements in AI explainability and control mechanisms to confidently integrate these powerful tools deeper into their CI/CD workflows.
Read original source