Foundry Local, part 3: getting the model ready
Download and prepare the approved model and execution provider before scanning files. A startup probe helps, but the parser and real workflow need their own checks.
Read article →Intune · Automation · Field notes
Intune automation, from real projects to practical lessons.
The architecture, code, trade-offs, failures and fixes behind automation built around Microsoft Intune, shared for the wider IT community.
- 30+
- open-source projects published on GitHub
- 9
- Intune and Azure automations documented here
- 0
- shared secrets left on endpoints
01 · Real projects
Born in the field
Every article starts from an automation that was designed, shipped and operated, not from a lab demo.
02 · Honest trade-offs
Failures included
Design decisions, constraints and the fixes that made a solution reliable, so you can adapt them.
03 · Secure by default
No secrets on endpoints
Certificates, managed identities and least privilege are the baseline of every design we describe.
Latest research
Recently published
Foundry Local, part 2: sorting files with local AI
When a filename tells us little, an opt-in local excerpt can help suggest a folder. Here is the code that checks the document, the answer and the allowed categories.
Read article →Foundry Local, part 1: how local AI works
What runs on the device with Foundry Local, how models get there, and where local folder suggestions end and OneDrive synchronization begins.
Read article →