AI agents are on track to become the primary way we interact with computers. But relying on plain text or voice makes complex tasks slow and often inaccurate. Agentic UI addresses this problem by allowing agents to generate user interfaces dynamically.

Agentic UI is no longer a pipe dream but is becoming reality. In this talk we will explore the new OpenAI App SDK and build Agentic UI components together on top of the .NET MCP server library. We will look at the new primitives that define this space and examine how MCP and UI components work together to create a new class of user experiences that combine both conversational and interactive components.

We will move beyond the proprietary OpenAI App SDK and also explore the new MCP Apps standard that was published as a draft in November 2025. This standard aims to define how Agentic UI components can be used across different agents such as ChatGPT, Claude, or Goose.

By the end of the session, you will understand not only how MCP and Agentic UI work but also how to build these systems yourself using open protocols designed for the next generation of AI-native apps. If you are interested in agent driven interfaces, next generation API integration, or the practical side of building AI powered user experiences, this talk will give you everything you need to get started.