They started using Puppet against Mac dev machines, then switched to internal tools I believe. But it was one of the first times it occurred to me that we should manage and use DevOps and SRE practices against user machines and user workloads where possible. The trouble is the tooling isn’t that mature yet, and we can’t assume real-time data or always connected machines, and we don’t have a herd because users only have one machine with them at any time and remote state deletion only works when you know you’re not deleting critical state, which in turn requires better tooling and a greater understanding of user application state persistence behaviour than most are willing to invest time in doing.

This is what makes Chromebooks so easy to maintain though: web apps and sandboxed Android apps can all easily sync to the cloud and compartmentalize their data.