Run an AI Agent on Azure in Minutes (OpenClaw + Azure OpenAI)

Description

#Azure #AIAgent #OpenClaw #LiteLLM #AzureOpenAI #Terraform #Packer #GPT4o *Run an AI Agent on Azure in Minutes (OpenClaw + Azure OpenAI)* Deploy a fully autonomous AI agent workstation on Azure using Terraform, Packer, OpenClaw, and Azure OpenAI. The agent runs on an Ubuntu VM with an LXQt desktop, backed by two models — GPT-4o and GPT-4o Mini — routed through a LiteLLM proxy. In this project we give the agent a single natural language instruction and watch it query Azure Cost Management, generate a styled HTML report, send it by email through Azure Communication Services, and schedule itself as a nightly recurring task — no scripts written by hand, no credentials managed, no additional configuration. WHAT YOU'LL LEARN • Deploying an AI agent workstation on Azure with Terraform and Packer • Routing Azure OpenAI models through LiteLLM for OpenAI-compatible access • Giving an AI agent access to Azure APIs through a VM managed identity • Configuring outbound email with Azure Communication Services • Driving real automation with plain English instructions INFRASTRUCTURE DEPLOYED • VNet with VM subnet and NAT gateway (East US) • Ubuntu 24.04 Azure VM (Standard_D4s_v3) with LXQt desktop and XRDP • Packer-built managed image with OpenClaw, LiteLLM, Chrome, VS Code, and developer tooling • LiteLLM proxy configured for GPT-4o and GPT-4o Mini via Azure OpenAI • VM system-assigned managed identity with Key Vault and Cost Management access • Azure Communication Services email with auto-verified managed domain • Azure Key Vault secrets for desktop password, OpenAI config, and email config GitHub https://github.com/mamonaco1973/azure-openclaw README https://github.com/mamonaco1973/azure-openclaw/blob/main/README.md TIMESTAMPS 00:00 Introduction 00:22 Architecture 00:50 Build the Code 01:05 Build Results 01:39 Demo