OpenClaw Setup Guide: Deploy Your Personal AI Assistant in 10 Minutes

OpenClaw is a self-hosted gateway that turns your favorite messaging apps into direct lines to a powerful AI assistant. Unlike hosted services, you control the infrastructure, the data never leaves your machine, and you can customize everything. This guide walks you through installation, configuration, and your first conversation.

What You'll Get

By the end of this guide, you'll have:

• OpenClaw Gateway running as a system service (launchd on macOS, systemd on Linux)
• WhatsApp or Telegram connected and ready to receive messages
• A personal AI assistant you can message from your pocket
• Full control over your data and privacy

Prerequisites

Before you begin:

• Node.js 22 or higher — check with node --version
• npm, pnpm, or bun — we'll use npm here
• API key from your chosen provider — OpenAI, Anthropic, or another supported model provider
• 10 minutes — seriously, that's all it takes

Recommended: Use the strongest latest-generation model available (Claude Opus 4.6, GPT-5.2, or equivalent) for the best experience and lower prompt-injection risk.

Step 1: Install OpenClaw

Open your terminal and run:

npm install -g openclaw@latest

Verify installation:

openclaw --version

You should see a version number like 2026.3.4 or similar.

Step 2: Run the Onboarding Wizard

OpenClaw includes a guided wizard that handles everything: workspace setup, model configuration, channel pairing, and daemon installation.

openclaw onboard --install-daemon

The wizard will walk you through:

Workspace setup — creates ~/.openclaw/ for config and credentials

Model provider selection — choose Anthropic, OpenAI, or another provider

API key entry — securely stored in ~/.openclaw/credentials

Channel pairing — connect WhatsApp, Telegram, Discord, or others

Daemon installation — installs the Gateway as a system service so it stays running

Choosing a Channel

The wizard supports multiple channels. Here's what to pick based on your use case:

• WhatsApp — best for personal use, requires QR code pairing
• Telegram — easiest setup, just create a bot via @BotFather
• Discord — great for communities and groups
• Slack — ideal for team workflows
• Signal, iMessage, IRC, Matrix, and more — all supported

For this guide, we'll use Telegram because it's the fastest to set up.

Pairing Telegram

During the wizard, select Telegram as your channel

Open Telegram and message @BotFather

Send /newbot and follow the prompts to create a bot

Copy the bot token (looks like 123456:ABC-DEF1234ghIkl-zyx57W2v1u123ew11)

Paste it into the wizard when prompted

The wizard stores your token securely and configures the channel.

Pairing WhatsApp

If you chose WhatsApp:

The wizard will display a QR code in your terminal

Open WhatsApp on your phone → Settings → Linked Devices → Link a Device

Scan the QR code

Done! OpenClaw is now connected to WhatsApp

Step 3: Start the Gateway

If you ran the wizard with --install-daemon, the Gateway is already running as a system service. Check its status:

openclaw status

You should see:

Gateway: running
Channels: telegram (connected)
Model: anthropic/claude-opus-4-6

If the Gateway isn't running, start it manually:

openclaw gateway --port 18789

Tip: The default port is 18789. You can change it in ~/.openclaw/openclaw.json under gateway.port.

Step 4: Send Your First Message

Now for the fun part. Open your chosen messaging app and send a message to your bot:

Example:

"Hey! What's the weather in San Francisco?"

Your assistant will respond within seconds. Try a few more:

• "Summarize the latest AI news"
• "Write a haiku about self-hosting"
• "Help me debug this Python error: [paste error]"

Step 5: Access the Control UI

OpenClaw includes a web dashboard for monitoring sessions, viewing logs, and managing config. Open your browser and navigate to:

http://127.0.0.1:18789/

You'll see:

• Active sessions — who's talking to the assistant right now
• Message history — full transcripts (stored locally)
• Gateway health — uptime, memory usage, model status
• Config editor — tweak settings without editing JSON files

Remote access: If you want to access the dashboard from another device, see the Remote Access guide. Tailscale Serve/Funnel is the recommended approach.

Step 6: Configure Your Assistant

OpenClaw stores its config in ~/.openclaw/openclaw.json. Here are a few settings you might want to tweak:

Set Your Timezone

{
  "agents": {
    "defaults": {
      "userTimezone": "America/New_York"
    }
  }
}

Add a Fallback Model

If your primary model is unavailable, OpenClaw can automatically fail over:

{
  "agents": {
    "defaults": {
      "model": {
        "primary": "anthropic/claude-opus-4-6",
        "fallbacks": ["openai/gpt-4o", "anthropic/claude-sonnet-4-5"]
      }
    }
  }
}

Restrict Who Can Message You

For security, you can allowlist specific phone numbers or usernames:

{
  "channels": {
    "telegram": {
      "allowFrom": ["@your_username", "1234567890"]
    }
  }
}

Tip: Use openclaw doctor to validate your config and catch common mistakes.

Next Steps

Add More Channels

You can connect multiple channels to the same Gateway:

openclaw channels login

Follow the prompts to pair WhatsApp, Discord, Slack, or any other supported channel.

Install Skills

Skills extend what your assistant can do. Browse the ClawHub marketplace:

npm install -g clawhub
clawhub search weather
clawhub install weather

Now your assistant can check the weather, manage GitHub issues, send emails, and more.

Enable Browser Control

OpenClaw can control a headless Chromium instance for web automation:

{
  "browser": {
    "enabled": true
  }
}

Restart the Gateway, and your assistant can now browse the web, fill forms, take screenshots, and more.

Set Up Heartbeats

Heartbeats let your assistant check in periodically without you prompting it. Useful for monitoring emails, calendars, or CI pipelines:

{
  "agents": {
    "defaults": {
      "heartbeat": {
        "every": "30m",
        "activeHours": {
          "start": "08:00",
          "end": "23:00",
          "timezone": "America/New_York"
        }
      }
    }
  }
}

Your assistant will wake up every 30 minutes during active hours and check for things that need attention.

Deploy to a Server

Running OpenClaw on your laptop works great, but deploying to a VPS means your assistant is always available. See the VPS Hosting guide for DigitalOcean, Hetzner, or Fly.io setups.

Troubleshooting

Gateway Won't Start

Check the logs:

openclaw logs --tail 50

Common issues:

• Port already in use: Change gateway.port in your config
• Invalid API key: Re-run openclaw onboard to update credentials
• Missing Node.js: Upgrade to Node 22+

Channel Disconnected

For WhatsApp:

openclaw channels login

Scan the QR code again.

For Telegram:

Verify your bot token is correct in ~/.openclaw/openclaw.json under channels.telegram.botToken.

Assistant Not Responding

Run a health check:

openclaw health --json

This shows model status, channel connections, and any errors.

Security Best Practices

OpenClaw is designed for single-user, personal use. Here's how to keep it secure:

Use DM pairing — set dmPolicy: "pairing" to require approval before unknown senders can message you

Enable disk encryption — FileVault (macOS), LUKS (Linux), BitLocker (Windows)

Run behind a firewall — don't expose the Gateway port (18789) to the public internet

Use Tailscale — for remote access, Tailscale Serve/Funnel is safer than SSH tunnels

Audit regularly — run openclaw security audit --deep to catch misconfigurations

What You Just Built

You now have:

• A self-hosted AI assistant that never sends your data to a third-party service
• Multi-channel access (message from WhatsApp, Telegram, Discord, wherever)
• Full control over the model, memory, tools, and behavior
• A foundation you can extend with skills, automation, and integrations

This is just the beginning. OpenClaw is a platform, not just a chatbot. Read the Architecture Guide to understand how sessions, agents, and tools work together. Explore the Skills system to add capabilities. Join the Discord to connect with other builders.

Welcome to the self-hosted AI revolution. 🦞