AI agents vs. Kubernetes: where each belongs in the control loop
The New Stack 1 day ago 16 ● 2 sources
Kubernetes has been evolving agent frameworks into its control-loop workflows, while platform teams are setting boundaries for what AI agents can change in live clusters. Over the past 18 months, agent tooling has moved from a side project into part of the cloud-native stack, with CNCF accepting agent frameworks into its Sandbox and SIG Apps taking on Agent Sandbox. The approach shifts to a split loop where agents handle investigation and proposal, while deterministic operators and policy/approval gates execute and verify changes.