AI workloads introduce a new kind of infrastructure problem. Demand can change rapidly, models have different compute and routing requirements, and traditional Kubernetes platforms still depend heavily on humans to translate those changes into infrastructure decisions.

What if the platform could close that loop itself?

In this session, we explore how Cluster API (CAPI) and Envoy AI Gateway can be combined to build the foundations of an AI-aware Kubernetes control plane. CAPI provides a declarative model for creating, scaling, and reconciling Kubernetes infrastructure, while Envoy AI Gateway provides an intelligent traffic layer for routing requests across models and AI workloads.

We will walk through an architecture where signals from AI workloads and gateway traffic can influence infrastructure decisions, creating a feedback loop.

Attendees will leave with an understanding of how CAPI and Envoy AI Gateway work, how they can be connected, where automation boundaries should exist, and what it would take to move Kubernetes platforms from human-operated infrastructure toward increasingly self-managing AI platforms