Platform engineering gave developers golden paths, self-service portals, service catalogs, and standardized Day-2 operations. Agentic engineering raises a new question: what happens when the “user” of that platform is not only a human developer but also an AI agent?

This talk explores how internal developer platforms need to evolve for agentic workflows: APIs rather than UIs alone, guardrails rather than tickets, policy-enforced automation rather than blind autonomy, and observable agent actions rather than invisible magic. We will look at what existing cloud-native patterns already solve, where agents introduce new risks, and how platform teams can prepare their Kubernetes and service platforms for AI-assisted software delivery without creating a new kind of operational chaos.