Empowering DevOps with OpenAI
AI is upending everything about the SDLC. For example, Microsoft has monetized GitHub Copilot, their AI development assistant, and now 40% of the code users check into GitHub is written by the AI. However, using Copilot is still highly dependent upon developer adoption and not fully integrated into the CI/CD pipeline and Paul wanted to bring AI into our agile process and DevOps environment. In this session, Paul Klinker will discuss how he and his team decided upon a phased approach for their customers, starting with integrating OpenAI into the CI/CD pipeline to enable greater visibility and understanding into code changes. Using the OpenAI API, he can explain, in plain English, what is happening when code changes. Project Managers, customers, product owners, and testers have less reliance on developers and can streamline the decision process for feature deployments, change impact assessments, determine the relevance of certain requirements and auto-generate release notes. Even other developers can benefit from this capability during source merges by providing insight into their colleague’s actions. The session will focus on the issues they encountered, including unexpected user behavior in the CI/CD pipeline, AI prompt engineering to get good explanations, institutional culture change to accept AI, and technical challenges in integrating with 3rd party tools.