Skip to content

Desktop, Server, or Core CLI: which to use

Wayland runs in three shapes. They share the same underlying agent engine and the same model providers, but each one serves a different context.

ShapeWhat it isBest for
Desktop appGraphical Electron app for Mac, Windows, and Linux.Day-to-day conversational work, teams, channels, GUI settings.
Server (self-hosted)Headless web server you deploy on a Linux box or VPS.Always-on access from a phone or any browser, shared household/team setups.
Wayland Core CLIRust CLI and embeddable engine (wayland-core).Scripting, CI, JSON stream embedding, terminal-native workflows.

The desktop app is the full graphical surface. It bundles the engine and wraps it in a point-and-click interface.

Use the desktop app when you:

  • Want a GUI for chat, assistants, and settings.
  • Are connecting messaging channels (Telegram, Slack, WhatsApp, and others). Channel pairing and routing live exclusively in the app.
  • Want to run multi-agent teams with specialist sub-panels.
  • Want the Scheduler, voice input/output, or image generation UI.
  • Want to reach your conversations from a phone through Remote Access.
  • Are not comfortable with a terminal.

The desktop app requires a supported OS (macOS, Windows, Linux) with a display. Download it at https://getwayland.com/download.

The server is the same agent packaged as the getwayland npm module and run headless on a Linux box or VPS. You reach it from any browser or phone by scanning a QR code printed in the terminal at first boot.

Use the server when you:

  • Want always-on AI without leaving your laptop open.
  • Want to share a single Wayland instance with family members or a small team, each with their own login.
  • Are deploying on a cloud VPS (DigitalOcean, Render, Hetzner, etc.).
  • Want to pair your phone and control Wayland without the desktop app.

The server needs a Linux x64 machine with roughly 2 GB of RAM and Node 18 or newer. It does not support macOS or Windows as a host. The web UI it serves is responsive, so it works on phones.

See Run Wayland on a Server for the three-step setup.

Wayland Core (wayland-core) is the open-source Rust engine exposed as a standalone command-line tool. It is the same engine the desktop app bundles, but without the GUI wrapper.

Use Wayland Core when you:

  • Are scripting Wayland in a shell or CI pipeline.
  • Want a one-shot prompt: wayland-core "summarize this file" < input.txt.
  • Are embedding the engine in another program via the JSON stream protocol.
  • Are running on a headless server with no browser and want a terminal UI.
  • Want to drive Wayland from code using the host-integration API.

Wayland Core is available via npm (@ferroxlabs/wayland-core), Cargo, and GitHub Releases. See Install Wayland Core (CLI).

CapabilityDesktop appServerCore CLI
Graphical chat interfaceYesWeb UI (browser)TUI only
Channel integrations (Telegram, Slack, …)YesPlannedNo
Multi-agent teamsYesPlannedVia CLI commands
Scheduler and cron UIYesPlannedwayland-core cron
Voice input and outputYesNoNo
Image generation UIYesPartialNo
Remote access (phone / browser)Via Remote Access featureNativeNo
Scripting / CINoNoYes
JSON stream embeddingNoNoYes
One-shot shell promptsNoNoYes
Headless Linux deploymentNoYesYes
Shared multi-user accessNoYesNo
Config file sharedYesPartialYes

Start with the desktop app if you are not sure. It is the most complete surface and requires no server setup.

Add the server if you want always-on access or want your phone to reach Wayland from anywhere.

Add the Core CLI when you have a specific automation job: a shell script, a CI step, or an embedding project.

All three can coexist. The desktop app, the server, and the Core CLI all read from compatible config formats, so a workflow you build in the desktop app can be driven from the CLI with the same provider keys.