CI/CD with Robert Erez

This conversation with Octopus Deploy engineer Robert Erez cuts through CI/CD dogma. The core insight is that most teams just want to ship software reliably, not adhere to ideological purity around GitOps or deployment strategies. Erez argues GitOps is a useful pattern, not a universal requirement—many teams cargo-cult it because conferences say so, then hit walls with secrets management, database migrations, or imperative workflow steps that don't fit the declarative model. The real trend is platform teams emerging as organizational scaffolding to reduce cognitive load on application developers who shouldn't need to become deployment pipeline experts. On AI's impact, Erez offers a contrarian early signal: if AI agents generate most code, pipeline speed may matter less than risk reduction. The bottleneck shifts from developer wait time to validation quality. Feature flags become more critical as the decoupling mechanism between deployment velocity and release safety. This isn't a tutorial—it's a field report from someone who has seen thousands of deployment patterns and knows that context collapse between conference best practices and operational reality is the real problem. Missing details include concrete metrics on AI-generated code volume or platform team adoption data. The evidence is primarily anecdotal from customer conversations but carries weight due to Octopus Deploy's broad enterprise exposure across regulated industries, retail, and even maritime research vessels.