From Control Planes to Cognitive Planes: Agentic Architecture with MCP

APIs helped define cloud-native—but agentic systems need more.

In this talk, we introduce MCP (Model, Context, Protocol)—an architectural framework built for adaptive, AI-enhanced, agentic systems. You'll learn how MCP extends API-centric thinking to define not just interfaces, but also capabilities, environment-aware behavior, and reflective discovery.

We’ll explore:

How MCP aligns with Kubernetes' dynamic, declarative model
Why JSON-RPC (not REST/gRPC) is the best fit for dynamic agent tooling

Real-world examples of LLM-driven clients discovering tools, prompts, and commands at runtime
How to structure MCP clients and servers using C# or Go for cloud-native deployment

You'll leave with code samples, a conceptual model, and a UML-driven design for building MCP-based systems—ideal for Kubernetes-based platforms, autonomous agents, and AI-enhanced -- developer tools.