An enterprise agent control plane needs five layers: identity, catalog, policy, telemetry and lifecycle. This is the operating model I would build.
Azure — Articles by Jannik Reinhard
Supported byAdvertisement
Five Gates Before an MCP Tool Reaches Production
MCP tool production needs more than a working connection. I explain five security gates for identity, scope, limits, approval and evidence.
Read article: Five Gates Before an MCP Tool Reaches Production
Azure API Management MCP: The Control Plane for Agent Tools
An MCP server is easy to connect, but an enterprise still needs identity, limits, monitoring and a clear tool boundary. In this blog post I use a live Azure API Management instance to show how I…
Read article: Azure API Management MCP: The Control Plane for Agent Tools
Microsoft Foundry Landing Zone: Governance in the Portal
Explore a Microsoft Foundry landing zone in the portal: shared services, identity, networking, model policies and the tests needed before a wider rollout.
Read article: Microsoft Foundry Landing Zone: Governance in the Portal
Multi-Agent Orchestration in Microsoft Foundry: A Deep Dive
In this blog post I explain how multi-agent orchestration works in Microsoft Foundry today. I show why connected agents and portal workflows are going away, how the Microsoft Agent Framework patterns work, and which orchestration pattern…
Read article: Multi-Agent Orchestration in Microsoft Foundry: A Deep Dive
Microsoft Foundry Model Deployment: A Practical Guide
A beginner-friendly Microsoft Foundry model deployment guide: choose a model, name the deployment and test the right fit before an app uses it.
Read article: Microsoft Foundry Model Deployment: A Practical Guide
Foundry IQ Deep Dive: Knowledge Bases for AI Agents
In this blog post I take a deep look at Foundry IQ, the knowledge layer in Microsoft Foundry. I explain how knowledge bases and agentic retrieval work, how to connect them to an agent via MCP,…
Read article: Foundry IQ Deep Dive: Knowledge Bases for AI Agents
Microsoft Foundry Setup: Project, Model and Deployment
A beginner-friendly Microsoft Foundry setup: create a project, understand models and deployments, then run one focused first test.
Read article: Microsoft Foundry Setup: Project, Model and Deployment
Microsoft Foundry Observability: Trace AI Agents in Production
In this blog post I explain how observability works for AI agents in Microsoft Foundry: zero-code tracing with OpenTelemetry, the monitoring dashboard, continuous evaluation on live traffic and alerts. With the exact setup I use in…
Read article: Microsoft Foundry Observability: Trace AI Agents in Production
Microsoft Foundry Realtime Voice Agent: Talk to Your Screen
Talk to live screen information with a Microsoft Foundry Realtime voice agent. Share an Azure page, ask by voice, and receive an immediate spoken answer.
Read article: Microsoft Foundry Realtime Voice Agent: Talk to Your Screen
Supported byAdvertisement