If your team is running AI coding agents, you're probably using VMs or Docker containers for sandboxing.
You should stop.
The VM tax on your team
Every AI task that needs a VM:
- 30 seconds to boot
- 2-4 GB of RAM allocated
- Networking setup
- File sharing setup
- Tooling installation inside the VM
- Cleanup when done
Multiply that by 10 tasks a day × 5 developers = a lot of wasted time.
The mapped drive alternative
On Windows, you can map any folder to a drive letter:
subst Z: C:i-workspaces\dev-1-task-1
Point the AI tool at Z:\. Done.
No VM overhead. No Docker learning curve. No networking setup.
Why this works for teams
- Each developer gets their own drives — no conflicts
- Each task gets its own drive — easy to review and clean up
- No resource overhead — just a folder on disk
- Works with every AI tool — Cursor, Claude Code, Cline, Aider, whatever
The tool
I use ScsSubst (https://subst.scsoi.com/) to manage these mapped drives. $9.99 lifetime, one-time purchase. It makes persistent mappings that survive reboots, and gives you a GUI to see all your drives at a glance.
When you actually need VMs
If you're running untrusted code from the internet, sure, use VMs. But if you're running your own AI coding assistant on your own machines, and you just want to make sure agents can't accidentally delete files — mapped drives are enough.
The bottom line
VMs are for running untrusted code. Mapped drives are for keeping your AI agent contained.
Use the right tool for the job.
Top comments (0)