Skip to content

Offline vs. Online

Flow-Like Desktop can create offline and online apps. This choice primarily controls persistence and collaboration; each Flow and Event also has execution settings that control where a run happens.

CapabilityOffline appOnline app
Primary app dataLocal deviceConfigured Flow-Like backend
Works without signing inYesNo
Available in the web appNoYes
Multi-device accessThrough explicit export/importThrough the online account
Team roles and invitationsNoYes
Publication workflowNoYes

Offline is a good default for personal experiments, local automation, and work that must stay on one machine. Online apps are the right choice when the app needs web access, collaboration, server-side events, or publication.

An offline app is not mutated in place. From its configuration, select Create an online copy to upload a new, secret-stripped copy to your account. The local source app remains unchanged.

The copy process removes secret variable defaults and known token fields before upload. Runtime values stored on the device are separate from the app bundle and are not used to configure the online copy. Review the new app’s Flows, storage, credentials, roles, and event settings before relying on it.

Online apps can likewise be forked to a local or another online destination when the source app and your permissions allow it.

The client and the Flow’s execution mode work together:

Flow modeFlow-Like DesktopWeb app or server-only caller
HybridNormally localRemote
LocalLocalNot remotely executable
RemoteRemoteRemote

An offline app runs on Desktop and cannot be invoked by the web backend. An online app can still run locally from Desktop when the Flow and required nodes allow it.

Hybrid means the same Flow may run locally in Desktop or remotely through the web/API path. It does not divide a single graph into local and remote sections.

Events have their own Local or Remote execution setting and reference either the latest Flow draft or a pinned Flow version. A remote event is useful for webhooks, schedules, and other triggers that must work while a user’s Desktop app is not running.

The selected Flow must be compatible with that location. Pre-run analysis can require local execution when a node needs capabilities unavailable on the server.

Local execution requires enough read access to obtain the Flow definition. A role that can invoke an app but cannot read its Flows uses server-side execution instead. Configure exact capabilities on the app’s Roles page.

Some nodes depend on the current device—for example desktop input, a local path, locally installed software, or attached hardware. When a Flow requires those capabilities, run it from Desktop and keep the app available on that machine.

A local run can still call remote APIs and cloud services. “Local” identifies the execution host, not a network-disconnected sandbox.

See Local-only execution and Runtime Variables for device-specific configuration.

With a self-hosted backend, remote runs stay on the infrastructure and execution backend you configure. They may use a shared runtime pool, a per-run Kubernetes Job, Lambda, or another supported execution backend; self-hosting does not imply one dedicated container per run.

Keep Desktop, web, API, and executor versions compatible when the same online app is used across them.