In this .NET Live session, Vic Perdana, Katie Savage, and Maira Wenzel reveal how to use Agent Framework and Azure AI Foundry to create practical, interoperable AI agents. The demo highlights SemantiClip—an open-source project that automatically turns video content into blogs.

Building Real-World AI Agents with Agent Framework on .NET Live

Presented by: Vic Perdana, Katie Savage, and Maira Wenzel

Overview

This session explores moving beyond AI hype to tangible developer outcomes using the new Agent Framework (AF). AF is the next-generation evolution of Microsoft's Semantic Kernel and Autogen, designed for building robust, composable, production-ready AI agents.

Featured Demo: SemantiClip

• SemantiClip is an open-source project that automates transforming video content into structured blog posts.
• Built using Agent Framework for modular, reusable agent logic.
• Integrates local and cloud-based LLMs.
• Demonstrates tool calling, AI-based planning, and persistent memory management with Model Context Protocol (MCP).

Technical Highlights

• **Agent Framework (AF)😗* The main focus, providing both ‘agent’ and ‘process’ frameworks:
  • Agent framework for smaller, focused co-pilot tasks
  • Process framework for orchestrating complex multi-agent workflows
• Azure AI Foundry: Used for deploying and scaling the solution with Microsoft cloud services.
• **Model Context Protocol (MCP)😗* Enables persistent, context-aware workflows by sharing memory/state across agents.
• Orchestration patterns: Best practices for interoperable, composable AI solutions

Core Topics Covered

• Real-world agent architecture for developers
• How to choose between different AF orchestration models
• Open-source patterns and practical lessons from building SemantiClip
• Integrating tool usage, LLMs, and memory with MCP
• Adapting the frameworks for co-pilots, automation, and multi-agent systems

Who Should Watch

• .NET and Microsoft cloud developers
• AI engineers interested in agent-based architectures
• Anyone seeking practical approaches to building and deploying AI experiences

Notable Takeaways

• How Agent Framework and Azure AI Foundry work together
• Patterns you can implement immediately for building composable, context-aware agents
• Key distinctions between orchestration models in AF

Resources

• Agent Framework documentation (if public)
• SemantiClip project repository (if public)
• Learn more about Azure AI Foundry
• Explore Model Context Protocol (MCP) for AI workflows