I'm Back — and the Lab Moved to the Cloud
It has been quiet here for a while. The most recent post in the archive is from October 2021, which is a rather long gap between updates. I thought it was time to explain what has changed and, more importantly, to start sharing some of the things I have been working on.
Some of you may remember my post about joining Microsoft in 2019. I started as a Premier Field Engineer and now work as a Cloud Solution Architect. The focus of my work has changed along the way, with more time spent on Azure Virtual Desktop, Windows 365 and Intune.
There have been changes outside work as well. My partner and I now have two children, and family life has understandably taken priority over writing blog posts. The gap was not a planned farewell to the blog. It simply became much easier to find something worth writing about than to find the time to write it.
Different tools, familiar problems
If you followed the original blog, you probably came here for MDT, Windows deployment, PowerShell or a particularly stubborn troubleshooting case. Those subjects had something in common: a practical problem, usually followed by an attempt to avoid doing the same work manually a second time.
That part has not changed. Building and maintaining a reference image is still a useful topic, even when the build runs in Azure rather than VMware Workstation. A configuration review still needs evidence, whether the settings come from Group Policy or Intune. Moving a desktop to the cloud has not made troubleshooting unnecessary either. Unfortunately.
The newer posts will reflect that work: image creation with Azure Image Builder, AVD assessments and session-host management, Windows 365, and the scripts that help with day-to-day endpoint operations. The aim is to explain the approach and the reasons behind it, including the parts you need to adapt to your own environment.
The code and the archive
The current scripts and tools are collected in endpoint-toolkit. Alongside PowerShell scripts, there are Bicep templates, Azure DevOps pipelines and desktop tools such as AvdAssessor, W365Assessor and BaselinePilot. The toolkit overview provides a starting point without having to work through every repository folder.
The older articles remain available in the archive and the Cases of the Unexplained section. I do not want to remove useful explanations just because the products have moved on. However, some articles reference an older code repository that I no longer maintain, and their instructions reflect the Windows versions available at the time. Please treat them as historical material, not current deployment guidance.
And a word about AI
AI-assisted work will be part of the discussion too. There are useful questions here: where can an LLM help with a script or an investigation, what information is appropriate to share with it, and how do you check the result?
I am interested in the practical answers. A plausible explanation still needs to agree with the logs, and generated code still needs to be reviewed and tested. Customer data also needs the same care as it would in any other tool. None of those responsibilities disappear because the first draft arrived quickly.
When I joined Microsoft, I wrote that sharing experience with the community was one of the most satisfying parts of this work. That is still the reason for this blog. A lot of what I know came from people who took the time to document a problem and explain their solution. I would like to continue doing the same.
As before, this is a personal blog, and the views expressed here are my own. Thanks for coming back. Now, about those images.