Source: vercel.com/blog/introducing-vercel-drives-persistent-storage-for-agents
Vercel says agents often run complex work across multiple runs and may need a new environment with different tools, runtimes, or compute. A drive keeps an agent's files independent of its execution environment, so a new sandbox can mount the same drive and pick up where the previous run stopped. The CLI and the Sandbox SDK both support creating a drive once and mounting it when a sandbox is created.
Drives can also be shared across sandboxes. One agent can prepare a dataset on a drive, and other agents can mount read-only copies to analyze it or generate reports without rebuilding it. Agents can also fork a drive, and each fork starts with the drive's contents and gets its own writable workspace. Forks do not merge back into the original drive.
Drives default to 1 TiB, or 1 GiB on the Hobby plan, with up to 16 TiB available. Documentation covers configuration, snapshots, and limits.
Why it matters if you build agents
Agent developers can move work between sandboxes or runtimes without copying the workspace, with files under the mount path persisting on the drive while everything else stays with the sandbox.
Forks do not merge back into the original drive, so teams need their own process for combining results from parallel experiments.
Get stories like this every weekday
One short email with the day's agent tooling news.