Crossplane and AI: The case for API-first infrastructure

AI-assisted development has shifted the bottleneck from writing code to everything that happens after git push — infrastructure provisioning, policy enforcement, and day-two operations. Most platforms are built for humans, not AI agents, which need unified, machine-readable APIs with explicit governance. Crossplane extends the Kubernetes declarative control model to all infrastructure and applications, exposing a single consistent API surface. Crossplane 2.0 allows composite APIs to provision infrastructure, deploy applications, configure networking, and define operational workflows in one place. Policy enforcement via RBAC, admission controllers, OPA, and Kyverno is architectural rather than procedural. This model lets AI agents declare intent and let controllers handle reconciliation, removing the need for tribal knowledge or manual coordination. The post argues that a declarative control plane is the prerequisite for AI agents to become first-class participants in infrastructure operations.