We've automated deployments and standardized governance. But getting from "a team needs a workload" to a governed, deployed environment is still largely manual: requirements workshops, architecture reviews, policy checks, IaC authoring, handoffs.

This session is a practitioner-focused deep dive into agentic platform engineering on Microsoft Azure: how to orchestrate AI agents that gather requirements, design a solution, plan and write IaC, deploy it, and document what was built, with a human approving every gate.

We'll start with a short map of the tooling available today: Azure Copilot, GitHub Copilot with Azure skills, and where each fits. Then we'll pivot to APEX (Agentic Platform Engineering eXperience for Azure), an open-source accelerator, and spend most of the session on how it works:
- The agent, subagent and skill model, and how a workflow graph with approval gates orchestrates them
- Guardrails by design: fail-closed adversarial reviews, a governance attestation chain from policy discovery to live precheck, and plan locking before deployment
- How APEX differs from general-purpose multi-agent frameworks such as Squad: a fixed, auditable workflow with human gates instead of free-form agent collaboration
- How it fits the Cloud Platform story and speeds up building workload landing zones on top of your existing IaC (Bicep or Terraform), GitOps and platform engineering practices

You will be able to take what you learned and start building tomorrow.