AI-500T00 - Design and Implement Multi-agent AI solutions Lab

This lab includes MS Learn content which is fully integrated in the account portal and lab environment. 

This lab builds the engineering skills needed to design, secure, operate, and govern production multi-agent AI systems on Microsoft Azure. Learners work through four synthetic customer scenarios, Contoso Capital, Northwind Health, Fabrikam, and Adventure Works, covering stateful agent loops, orchestration and task decomposition, A2A communication, advanced prompting, MCP tooling, RAG, shared memory, CI/CD, zero trust, responsible AI, lifecycle governance, observability, evaluation, cost optimization, human approval, and incident response. Every lab runs from Visual Studio Code with Python, the Azure CLI, the Azure Developer CLI, and Bicep, and provisions live services such as Microsoft Foundry Agent Service, Azure Cosmos DB, Azure AI Search, Azure Container Apps, Service Bus, Event Hubs, Application Insights, and Azure Managed Redis. Security and privacy run through the whole course rather than sitting in one module, with Microsoft Entra authentication in place of keys, tenant and patient isolation, synthetic data only, and minimized telemetry. Each lab closes with explicit validation against live evidence, typically Foundry traces, Cosmos DB Data Explorer items, KQL query results, or saved JSON artifacts, followed by a full cleanup of billable resources.

This lab includes a Cloud Express Pass (CEP), which provides access to the cloud resources required to complete your hands-on lab exercises. Your CEP is included with your lab purchase and can be found under the Info tab after launching your lab.

Labs are available in English.