Microsoft Ignite 2026 Day 1: My Intune and Device Management Roundup

It’s been a while since I posted, perhaps going hard in early October. But things have been hectic… and now, things are about to get a bit more hectic. This week we’ve got Ignite, then at the end of the week the long awaited GTA VI release. I’m not really a gamer, but still, this is one you simply cannot miss I feel so will give it a poke. Conscious that events like Ignite can quickly pass you by if you don’t dive in, I’ve put a lightweight roundup together focusing solely on Microsoft Intune and device management. It covers (most of) what was announced on 17 November, what I think deserves attention, and what I’d actually be prepared to play with myself.

There’s a fair amount of “preview” talk in this list as you might expect, including private and limited previews. So we’ll not be able to get our hands dirty immediately, but this highlight things to research, investigate and evaluate; something announced doesn’t automatically give me something I can deploy across a production estate this afternoon.

Windows Server management is coming to Intune

Microsoft is extending Intune management to eligible Windows Server Insider builds, with supported configuration, compliance, application and update workflows. It’s a non-production preview, intended for a separate test environment. Server Core, multisession and nonpersistent hosts are outside its current scope. Microsoft’s server-management article explains the boundaries.

If I was exploring this, my first question would be about ownership. Who manages these servers today, and what problem would bringing them into Intune solve? I’d want a clear answer before worrying about the enrolment steps.

I’d also review broad assignments, filters and administrative access before the first server enrols. Microsoft calls out those checks explicitly. A policy that makes perfect sense for a laptop deserves another look before it reaches a server. Familiar tooling is useful; assuming every endpoint configuration is appropriate would be rather less useful.

Azure Arc remains relevant to infrastructure management. I would map responsibilities between tools rather than turn this into a competition over which console gets to own everything.

Local AI agents get more specific Intune controls

Software compliance is in public preview; MXC containment policies are rolling out. The distinction is important. Software compliance can use supported application identifiers to mark a Windows device noncompliant, with Conditional Access governing access to protected resources. It doesn’t identify every possible extension, package or standalone agent.

For approved, supported MXC-enabled agents, Intune’s containment policies define file and network boundaries. Initial coverage includes process containers and supported agents beginning with GitHub Copilot CLI and OpenAI Codex CLI. The 17 November governance guide covers these controls and their limitations.

For me, this is the practical follow-on to discovering local AI agents. Once I know what’s present, I need a decision: approve it, investigate it, restrict access or define the resources it needs. I’d keep that decision tied to a business task, with a named owner.

I’d be especially careful with the compliance route. Restricting corporate access for the device has consequences for the employee’s other work. My pilot would include the communication and recovery experience, not just proof that access gets blocked.

Application management: promising changes, with preview labels attached

Microsoft’s Day 1 Intune announcement identifies device-level app inventory as public preview. EAM automatic updates for more eligible Windows apps, including some Intune didn’t deploy, and app-in-use detection are private previews.

I’d put these on the evaluation list together. Better inventory helps identify the work; updating applications still needs a sensible approach to compatibility and interruption. My test would include an application somebody is actively using, with unsaved work and a business deadline. That’s where deployment behaviour becomes very real.

I’d want the application owner involved from the start. Agreeing the permitted disruption beforehand is much easier than explaining it afterwards.

CIS auditing builds on the assessment story

CIS benchmark auditing is announced for public preview this November in Advanced Analytics, alongside the existing STIG audit approach. It assesses configuration without changing device settings.

That gives me another reason to be precise about what “baseline” means in a project conversation. I’d use an assessment to identify gaps, agree exceptions and plan remediation. I’d still expect the actual configuration changes to have owners, testing and a record of why they were made.

More visibility during Autopilot device preparation

A limited public preview adds a Company Portal experience after OOBE, showing application installation and compliance progress for supported deployments.

I like the focus on the employee here. I’d test whether someone can understand what is ready, what is still happening and when they should ask for help. A technically successful build can still generate a support call if the person receiving it has no idea what to expect.

AI-assisted investigation and remediation inside Intune

Microsoft Perception’s Intune investigation experience and the Intune Vulnerability Remediation Agent are limited public previews. They bring conversational investigation and remediation planning into endpoint operations.

I’d evaluate the evidence behind the recommendation, the proposed target population and the control over execution. My acceptance criteria would include an administrator being able to explain the proposed action before it happens. I’d also want to compare the result against the original question, rather than judge the experience by how confidently it answers.

What I’d prioritise after Day 1

My shortlist would be a separate Windows Server lab, a practical review of local-agent controls, and an application-update pilot when access to those previews becomes available. Hardly a shortlist when you’ve called out a full blown Server lab, but hey ho. Shortlist/wishlist, it may never happen.

U;ltimately, there’s plenty here to investigate. What I want from the next stage is straightforward: evidence that the new capabilities solve a defined management problem, fit the support model and behave predictably on the devices we’re responsible for, otherwise, probably one for the bin for me…

James avatar

Leave a Reply

Your email address will not be published. Required fields are marked *