KubeCon + CloudNativeCon India 2026

AI Agents are moving from experimental scripts to autonomous "digital employees" creating a new security crisis. Because agents determine their own execution paths, standard security fails. Securing them requires strictly enforcing the 4 Pillars of Governance: Administration, Authentication, Authorization, and Audit at the infrastructure layer.

This talk presents a practical implementation of these pillars using Cell-Based Architecture on Kubernetes. We will demonstrate how to encapsulate agents into governed "Cells" to enforce Zero Trust controls at every critical traffic boundary:
1. User → Agent
2. Agent → Agent
3. Agent → Tool
4. Agent → LLM

We will translate these pillars into real Kubernetes primitives, giving attendees a blueprint to move from "unmanaged bots" to hardened, enterprise-ready workloads. All patterns demonstrated use CNCF-hosted projects and open-source identity primitives leveraging open standards like OIDC and OAuth 2.0.