While many organizations are experimenting with AI agents, very few successfully scale them into production. Developers often have to choose between visual low-code tools that have a defined ceiling, and complex pro-code frameworks. Furthermore, monolithic agents frequently fail or hallucinate when handling complex enterprise tasks. In this session, we explore the shift towards specialized multi-agent architectures using the new Microsoft Agent Framework. We will demonstrate how a hybrid approach combines the best of both worlds.

Key Takeaways
• When to rely on low-code and when pro-code (Agent Framework) is strictly required
• Best practices for orchestrating multiple agents for complex enterprise workflows
• Governance, security, and tracing of agents in Azure AI Foundry

Target audience:
AI Engineers, Architects and Developers

Session duration:
45–60 minutes

Products:
Microsoft Agent Framework, Azure AI Foundry, Copilot Studio