Sync that stays out of the way
The desktop agent holds one connection open and streams changes as they happen. Large files transfer in chunks and resume where they stopped, so a dropped link costs seconds rather than a restart.
History you can actually use
Every write creates a version. Restore a single file, a folder, or the whole workspace to any point inside your retention window — from the web console or the CLI.
An endpoint, not a silo
The same buckets are reachable over an S3-compatible API. Existing backup jobs,
build pipelines and rclone configurations work without modification.
How it fits together
Clients connect to the endpoint nearest them and keep that connection open for the life of the session. Writes land on a replicated primary volume and are versioned on the way in; where the plan includes it, a copy goes off-site overnight.
Retention, in plain terms
Recent versions are kept in full. Older ones are thinned on a fixed schedule so the history stays useful without growing without bound. The retention window is set per workspace and can be changed at any time; lengthening it applies from that day forward.
What you get on every plan
| Transport | TLS 1.3 only. Older versions are refused, not downgraded. |
|---|---|
| At rest | AES-256 per volume, keys held on the storage nodes. |
| Durability | Erasure coded across three nodes. A nightly copy to a second region is included on Orchard and available on the other plans as an add-on. |
| Maximum object | 5 TB, uploaded in resumable parts. |
| Clients | macOS, Windows, Linux, iOS, Android, and a static CLI binary. |
| API | S3-compatible: ListObjects, multipart upload, presigned URLs. |