Building Secure Multi-Agent Workflows with Foundry Control Plane and GPU Agents [AMER/EMEA]

Description

*Brought to you by Microsoft and NVIDIA* GPU-accelerated agents enable advanced AI workflows, but coordinating them securely and reliably requires clear separation between control and execution. In this joint session, Microsoft and NVIDIA present a hybrid multi-agent architecture where Foundry Agent Service coordinates GPU-backed agents running on Azure Container Apps. Using a document processing scenario with multiple open-source models including NVIDIA Nemotron models, we show how Foundry agents route requests, apply security controls, and monitor execution across GPU workloads, producing consistent and explainable results suitable for production environments with NVIDIA serverless GPUs. We’ll demonstrate how to secure, observe and control agent behavior end-to-end using Foundry Control Plane including lifecycle visibility, policy enforcement, auditability, and inter-agent communication patterns. The sample scenario features a Stable Diffusion-based image generation agent on Azure Container Apps called by a Foundry-hosted coordinator agent. If you’re building multi-agent systems in production, this session shows how to keep them reliable, governed, and enterprise-ready. 📌 This episode is a part of a series. Learn more: https://aka.ms/AIAgentsApps/y-MSFT #MicrosoftReactor #LearnConnectBuild [eventID:26660]