OpenClaw Onboarding

The Five-Step Onboard Flow

1. Launch — start a container (self-host or managed)
2. Pair — approve your device via a 6-digit code
3. Model — pick the LLM your agent uses
4. Channel — connect Telegram, Discord, WhatsApp, WeChat, or web chat
5. Chat — send your first message and verify the round-trip works

Step 1: Launch

The fastest path is OpenClaw Launch: sign up, click Deploy, and a container is yours in roughly 30 seconds. Self-hosting works too — see the install guide for Docker, npm, and binary options.

Managed onboard sets these defaults for you:

• Gateway on *.openclawlaunch.app with valid TLS
• Pairing mode enabled (no static tokens to copy)
• OpenRouter wired up so every model is one click away
• Persistent workspace mounted at /home/node/.openclaw

Step 2: Pair Your Device

When you open the gateway URL, you see a 6-digit pairing code. Approve it from your account dashboard and the gateway issues a device-scoped token. This replaces the fragile copy-paste-a-secret flow from earlier OpenClaw versions.

If the pairing screen never appears, the most common cause is a misconfigured reverse proxy — see the pairing fix guide.

Step 3: Pick a Model

Onboard ships with a default model so you can chat immediately, but the first thing most people change is agents.defaults.model.primary. Free trial deploys default to a fast budget model; paid plans can switch to Claude, GPT-5, Gemini 2.5, DeepSeek, or any OpenRouter ID.

{
  "agents": {
    "defaults": {
      "model": {
        "primary": "openrouter/anthropic/claude-sonnet-4.6"
      }
    }
  }
}

Model edits are hot-reloaded — no gateway restart, no dropped sessions.

Step 4: Connect a Channel

Every channel needs both halves wired in openclaw.json:channels.X.enabled AND plugins.entries.X.enabled. Set only one and the channel registers silently and receives nothing.

Common channel guides:

• Telegram — use dmPolicy: "pairing"
• Discord — use dmPolicy: "open"
• WhatsApp — QR pairing
• WeChat — QR + plugin install on first connect

Step 5: First Chat

Open the gateway web UI or your channel and send any message. The first response is the slowest one (the gateway has to lazy-import the provider SDK and warm a sync mirror for skills). Subsequent messages stream back in under a second.

Managed deploys on OpenClaw Launch pre-fire that first chat at warm-pool creation, so users hit a hot gateway from the first message.

Onboard Checklist

Skip the Setup

OpenClaw Launch wires every step above by default. Plans from $3/mo first month, then $6/mo Lite or $20/mo Pro.