Real-world multi-agent implementations often reveal hidden challenges that can make or break your solution. This session bridges this gap by exploring both the depth of orchestration mechanics and the breadth of ecosystem integration possibilities.

What makes this different? We’ll tackle real challenges based on our experience in deploying these multi-agents solutions. Examples that are specific and that you won’t find in the official docs.

The Depth Journey
Discover the architecture behind the choice between connected agents vs. child agents and why it matters. Learn how to maintain context across multi-step workflows without bouncing back to the orchestrator unnecessarily. Understand proactive messaging capabilities and why connected agents can’t ask follow-up questions without a direct topic mandate – and what to do about it.

The Breadth Expedition
Uncover how to extend Copilot Studio beyond native capabilities by integrating Azure AI Foundry agents, advanced AI models, and custom AI connectors. See a complete end-to-end workflow demonstrating multi-model orchestration with specialized operation and communication agents.

Through two complementary real-world scenarios you’ll gain actionable insights into building enterprise-grade multi-agent solutions that actually work in production.

Key Takeaways:

• Understand the critical differences between connected and child agents for context retention

• Learn patterns for maintaining conversation ownership in multi-step processes

• Discover how to integrate third-party AI models seamlessly

• See practical implementations of specialized agents for operations and communications

• Master the art of orchestrating multiple AI platforms within a single workflow

This isn’t another “how to create your first agent” session – we would require a bit of prior experience thus we advice this session to practitioners ready to build sophisticated, production-grade solutions.