AI agents are rapidly moving beyond simple conversational chatbots to autonomous systems capable of observing environments, reasoning over complex telemetry, and executing remediation actions. But what happens when an AI agent is given operational responsibilities over production Kubernetes workloads?
This session explores how autonomous AI agents can safely operate and maintain Kubernetes workloads at the edge and in disconnected environments. Using a local LLM, Kubernetes, the Model Context Protocol (MCP), OpenTelemetry, and strict Kubernetes RBAC, we demonstrate an end-to-end self-healing cluster workflow: detecting a workload failure, investigating logs and cluster events, diagnosing the root cause, and executing controlled, policy-guarded remediations.
Attendees will discover how combining cloud-native observability with robust Kubernetes security primitives provides the indispensable guardrails required for AI agents to operate reliably without human intervention.
Key Takeaways
- Autonomous Cluster Remediation: See how AI agents leverage local LLMs and tools to diagnose incidents and perform automated root-cause analysis on Kubernetes clusters.
- Model Context Protocol (MCP) in Ops: Learn how to bridge Kubernetes APIs, runtime diagnostics, and LLMs using MCP for modular operational tool execution.
- Observability as Context: Discover how OpenTelemetry metrics, traces, and Kubernetes events provide high-fidelity grounding context for operational agents.
- Guardrails & Security: Understand how Kubernetes RBAC, admission controllers, and least-privilege security policies prevent agent hallucinations from impacting production stability.
Target Audience
- DevOps engineers, SREs, and Platform Engineers managing Kubernetes clusters and edge deployments.
- Cloud-native practitioners interested in integrating AI agents and autonomous self-healing into infrastructure workflows.
- Developers and architects exploring Model Context Protocol (MCP) and LLM-driven operational automation.
Prerequisites
- Basic understanding of Kubernetes concepts (Pods, Deployments, Logs, RBAC).
- Familiarity with containerization and general observability concepts.
