Your Agent Has No Idea Where It's Running

AI agents are being deployed at scale without any awareness of the infrastructure beneath them. They assume a flat, trusted execution environment. They're wrong.

The cloud-native community spent a decade solving exactly these problems: workload identity, service mesh, secure sandboxing, observability across distributed systems. We built Kubernetes so workloads wouldn't need to know where they were running. We built mTLS so services could trust each other without trusting the network. We built gVisor and Kata Containers so untrusted code couldn't escape its environment.
Then we handed the keys to AI agents and forgot to tell them any of it.

This talk is about the gap and why closing it matters more now than it ever did for microservices.