Beyond Chat: Bringing Agents To Kubernetes
summary
Kagent is a Kubernetes-native, declarative agent runtime that turns CRDs into agents, model configs, and MCP server tools; it launches per-agent Deployments, stores chat sessions in a relational DB, and emits logs/traces for observability. Demos show a GitHub-backed MCP server creating a PR to fix a wrong targetPort (needed a retry because model temperature caused a flaky fix) and a Cilium policy patch that updated labels to restore connectivity. Supports up to 20 tools, A2A nested agents, BYO-agent images via the ADK, and event-driven invocations via khook.
AI-written summary · about
chapters (9)
- 0:00Introduction and motivation
- 3:00Introducing Kagent
- 4:53GitOps demo and agent basics
- 9:43Agent core components
- 11:11Kagent deep dive
- 14:00Evolution and BYO agents
- 16:53Cilium policy demo
- 22:19Best practices and observability
- 24:31Conclusion and Q&A
related talks
23mAgents in Kubernetes 101: A Crash Course on Cloud Native AI Agents
19mScaling Agents on Kubernetes with acpx and ACP
27mFrom Chatbots to Agents: Orchestrating Agentic AI on Existing Enterprise Kubernetes Infrastructure.
29mBuilding Interoperable Agents: When A2A meets MCP
40mAgents in Action: Advancing Open Source AI Missions
36mPutting AI Agents to Work: A New Era of Open Source Connectivity with MCP
35mBuilding Agentic-Ready Infrastructure with kcp Workspaces and API Export
01h 42mHow to build and scale multi-agent AI systems