LLMs and AI agents are running in production, trained and served on Kubernetes. This newsletter is about running it all well: model inference and training, agent infrastructure, and cloud-native systems.