Fundamentals

Disposable VM vs. Snapshot: What's the Difference?

The conceptual answer is that a snapshot preserves state on purpose, and a disposable environment discards it on purpose. This is the concrete version — what Velo Workspaces' own restore points actually copy on disk, and what "disposable" actually does instead, verified against the app's own source rather than asserted.

What a restore point actually copies

A Velo restore point is a real, literal copy: creating one copies the workspace's primary disk image, and — if the workspace was hibernated rather than just powered off — its saved VM state too, into a dedicated snapshot folder on disk. Restoring one reverses that copy, bringing the workspace's disk (and saved state, if the snapshot has one) back to exactly what it was at that moment. Restoring isn't destructive to fail into, either: the app backs up the workspace's current disk and state before overwriting anything, and rolls that backup back into place automatically if the restore itself hits an error partway through.

What a disposable workspace does instead

Nothing is copied or preserved at all — that's the entire point. Closing a Disposable Workspace deletes its bundle outright: the disk, any state on it, gone. There's no snapshot involved anywhere in that path, because a disposable workspace was never meant to be returned to. The next one you create doesn't restore from the one you just closed; it clones fresh from a base image, a separate, sealed artifact that was never part of the workspace you just threw away — see base images as versioned infrastructure for that half of the mechanism.

Why you'd reach for each

A restore point is for a single working session you want to be able to rewind — take one before trying something risky (a migration, a config change you're not sure about), and if it goes wrong, restore rather than starting the whole environment over from scratch. Disposability is for the opposite goal: guaranteeing nothing from this run carries into the next one, which matters most exactly where the disposability post covers it — testing, QA, and anywhere an AI agent's generated code is running.

Not mutually exclusive

These aren't competing features you pick one of forever — they answer different questions, and a single Mac's set of workspaces can use both at once. A long-lived project workspace might lean on restore points as checkpoints during a risky refactor, while a completely different, disposable workspace handles the next CI run with no snapshot involved at all. Which one a given workspace needs depends on whether you want to be able to go back to it, not on some general rule about which pattern is "better."