Migrating an On-Premises AI Agent to Microsoft Azure Virtual Machine

Overview

VNB Consulting migrated an established OpenClaw AI agent from a locally hosted macOS environment to an Azure Virtual Machine. This on-premises AI agent migration to Microsoft Azure established a centralized cloud environment while preserving the agent’s workspace, memory, prompts, skills, scheduled processes, integrations, and Anthropic-based AI capabilities. The Azure-hosted agent retained its core business capabilities and connectivity with Anthropic and Microsoft 365, including the Microsoft Teams bot app and Outlook.

Customer Profile

The customer manages residential communities across the United States, coordinating leasing, resident support, property operations, maintenance, owner services, and stakeholder communications. Its business connects the people, services, and day-to-day activities involved in supporting well-managed residential communities.

Customer Situation

OpenClaw AI agent running locally on an on-premises environment.

The customer’s OpenClaw AI agent ran in a locally hosted macOS environment to support business automation and reporting. As the agent became increasingly important to operations, the customer sought to move it to a centrally hosted Microsoft Azure environment and enable its integration with Microsoft Teams bot app and Outlook.

The agent relied on interconnected workspace data, configurations, prompts, skills, memory, session history, scheduled processes, runtime dependencies, plugins, and external integrations. The customer therefore needed an approach that would:

  • Preserve the agent’s purpose, business context, prompts, knowledge, skills, and workflows.
  • Transfer its OpenClaw workspace, memory, session data, configurations, scheduled processes, and runtime dependencies.
  • Maintain authentication and connectivity with Anthropic and Microsoft 365 (Microsoft Teams bot app and Outlook).
  • Retain Microsoft Graph integration and Outlook email draft creation.
  • Confirm that the migrated agent could run its core business workflows in Azure.
  • Provide a controlled path for user acceptance testing and production cutover.

Our Solution

OpenClaw AI agent running on Microsoft Azure virtual machine.

VNB Consulting assessed the existing OpenClaw environment and designed a migration approach centered on a Microsoft Azure Virtual Machine . This cloud-hosted approach retained the OpenClaw runtime and existing Anthropic model access while minimizing changes to established workspace and file-based processes.

Our approach included:

  • Discovery and runtime assessment: Reviewed OpenClaw configurations, workspace data, session history, runtime dependencies, Python scripts, scheduled jobs, prompts, skills, tools, plugins, workflows, and integrations to identify dependencies and validation requirements.
  • Azure Virtual Machine deployment: Deployed OpenClaw to an Azure Virtual Machine running Ubuntu 24.04 and validated agent startup, scheduled execution, report generation, and connectivity with Microsoft Graph, Microsoft Teams, and Outlook.
  • Agent workload migration: Transferred the workspace, memory, session data, configurations, scheduled processes, runtime dependencies, and integration settings while preserving the agent’s behavior and business context.
  • Anthropic and Microsoft 365 integration: Retained Anthropic model access and Microsoft 365 authentication, and validated Microsoft Graph connectivity, Outlook draft creation, and core workflows.

Business Outcomes

  • Operational continuity: Preserved workflows, business context, workspace data, and integrations during the migration.
  • Cloud-hosted runtime: Shifted the agent from local macOS hosting to a centralized Microsoft Azure environment.
  • Infrastructure modernization: Modernized the agent’s hosting environment through an Azure VM without redesigning the existing solution.
  • Microsoft 365 connectivity: Allowed the Azure-hosted agent to authenticate with Microsoft 365, use Microsoft Graph, chat within Microsoft Teams bot app, and create Outlook email drafts.
  • Preserved AI investment: Retained the customer’s existing Anthropic model access without changing AI providers.

Related Case Studies

Property Management Organization Advances Customer Communication Through AI-Powered Email Response Automation with Microsoft Foundry

VNB Consulting helped a property management organization build an AI agent to prepare knowledge-grounded email responses with human review and approval.

VIEW DETAILS