From AgentKit to Azure Meltdown: Top Stories for Microsoft IT Pros

This Week in IT

This Week in IT

LISTEN ON:

OpenAI just dropped a toolkit for building AI agents—AgentKit. It promises to revolutionize the use of AI agents in the workplace. Also this week, Microsoft quietly announced it will start moving your oldest emails into the archive but then changed their minds. And a Kubernetes crash knocked out Azure Front Door and Microsoft Entra, leaving admins scrambling.

Links and resources

Episode overview

The top stories in this week’s episode:

OpenAI Unveils AgentKit for Enterprise AI Agents

OpenAI has launched AgentKit, a developer-focused toolkit designed to simplify the creation and deployment of AI agents in the workplace. AgentKit features a visual workflow canvas, a centralized connector registry, ChatKit for natural language processing, and evaluation tools for agent performance. Unlike Microsoft’s Copilot Studio—which is aimed at business users and tightly integrated with Microsoft 365 security—AgentKit is model-agnostic and highly flexible, allowing integration with any large language model and external workflow engines like Temporal. Some analysts are calling this OpenAI’s “iPhone moment” for enterprise AI, predicting it will accelerate reliable adoption of AI agents in business operations.

Microsoft’s Exchange Online Archiving Policy Reversal

Microsoft announced a controversial change: old emails in Exchange Online would be automatically archived once mailboxes reached 90% capacity. The move was prompted by the surge in AI-generated content filling inboxes. After pushback from businesses, Microsoft reversed the decision and will review the timeline for enabling this feature by default, giving organizations more time to prepare.

Azure Outage Highlights Cloud Resiliency Concerns

A major crash in Kubernetes disrupted Azure Front Door and Microsoft Entra, causing widespread outages for Microsoft cloud services—especially in Europe, the Middle East, and Africa. Microsoft engineers restored services by rebooting servers, but the incident has reignited concerns about the reliability and resiliency of cloud infrastructure. The episode serves as a reminder that while cloud providers handle recovery, organizations must plan for such disruptions in their own IT strategies.