How to Set Up OpenClaw: Complete Beginner's Guide (2026)

Getting Started · 15 min read

How to Set Up OpenClaw: The Complete Beginner's Guide (2026)

Last updated: February 2026 · Reading time: 12 minutes · Difficulty: Beginner-friendly

OpenClaw is a personal AI assistant that lives inside the messaging apps you already use — WhatsApp, Telegram, Slack, Discord, and more. It can handle your emails, manage your calendar, send reminders, browse the web, and automate the tedious tasks that eat up your day.

The catch? You have to set it up yourself. And if you're not a developer, that can feel intimidating.

This guide walks you through every step. No coding experience needed. By the end, you'll have your own AI assistant running on your computer, ready to chat with you on WhatsApp or Telegram.

What You'll Need Before You Start

Before we dive in, make sure you have these three things ready:

A computer that stays on (or a small server). OpenClaw runs as a background process, so it needs a machine that's always available. A Mac, Windows PC, or Linux box all work fine. If you want it running 24/7 without leaving your laptop open, a cheap cloud server works too — more on that later.

A messaging account. WhatsApp, Telegram, or any of the supported channels. Telegram is the easiest to set up because it uses a simple bot token. WhatsApp works great too but requires scanning a QR code to link your device.

An Anthropic API key (or OpenAI key). OpenClaw uses Claude or GPT as its brain. You'll need an account with Anthropic or OpenAI and an API key. The cost is usage-based — most people spend $3–10 per month depending on how much they chat.

[IMAGE: Clean graphic showing the three prerequisites — computer icon, messaging app icons, API key icon. Simple, not cluttered.]

Step 1: Install Node.js

OpenClaw is built on Node.js, so you'll need version 22 or newer installed on your machine.

On Mac:

Open Terminal (search for "Terminal" in Spotlight) and paste this command:

curl -fsSL https://fnm.vercel.app/install | bash

Close and reopen Terminal, then run:

fnm install 22
node --version

You should see something like v22.x.x. If you do, you're good.

On Windows:

Windows users need WSL2 (Windows Subsystem for Linux) first. Open PowerShell as Administrator and run:

wsl --install

Restart your computer, then open the Ubuntu terminal that appears in your Start menu. From there, follow the same Mac instructions above.

On Linux:

Same as Mac — the fnm commands work identically in your terminal.

[IMAGE: Screenshot of terminal showing successful Node.js installation with version number visible]

Step 2: Install OpenClaw

With Node.js ready, installing OpenClaw is one command:

npm install -g openclaw@latest

This downloads and installs the latest version globally on your machine. It usually takes about a minute.

To verify it worked:

openclaw --version

You should see the current version number printed back.

[IMAGE: Screenshot of terminal showing the npm install command completing successfully, followed by the version check]

Step 3: Run the Onboarding Wizard

This is where the magic happens. OpenClaw includes a built-in wizard that walks you through configuration step by step:

openclaw onboard --install-daemon

The --install-daemon flag is important — it sets up OpenClaw to run automatically in the background so it starts up again if your computer restarts.

The wizard will ask you a series of questions. Here's what to expect.

Choosing Your AI Model

The wizard asks which AI model you want to power your assistant. Your main options are:

Anthropic Claude (recommended) — Claude Opus 4.6 is the default and works best for long conversations and complex tasks. You'll need an Anthropic API key, which you can get at console.anthropic.com.

OpenAI GPT — If you already have an OpenAI account, this works too. The experience is slightly different but still excellent.

When prompted, paste in your API key. The wizard stores it securely in your local configuration file at ~/.openclaw/openclaw.json.

[IMAGE: Screenshot of the onboarding wizard running in terminal, showing the model selection step]

Connecting a Messaging Channel

Next, the wizard helps you connect your first messaging channel. We'll cover the two most popular options.

Step 4: Connect WhatsApp or Telegram

Option A: Telegram (Easiest)

Telegram is the simplest channel to set up because it uses a bot token — no phone number linking required.

Create a Telegram bot:

  1. Open Telegram and search for @BotFather
  2. Send /newbot
  3. Choose a name for your bot (this is the display name, like "My Assistant")
  4. Choose a username (must end in "bot", like myassistant_bot)
  5. BotFather sends you a token — it looks like 123456789:ABCDefGHIjklMNOpqrSTUvwxYZ

Add the token to OpenClaw:

The onboarding wizard will ask for this token. Paste it in when prompted. If you've already finished the wizard, you can add it manually to your config file:

{
  "channels": {
    "telegram": {
      "botToken": "YOUR_TOKEN_HERE"
    }
  }
}

Now open Telegram and message your bot. It should respond.

[IMAGE: Screenshot of Telegram conversation with BotFather showing the bot creation process and token]

[IMAGE: Screenshot of first successful message exchange with the OpenClaw bot in Telegram]

Option B: WhatsApp

WhatsApp requires linking your phone, similar to WhatsApp Web.

Run the channel login command:

openclaw channels login

A QR code appears in your terminal. Open WhatsApp on your phone, go to Settings → Linked Devices → Link a Device, and scan the code.

Once linked, set up who can message your assistant by editing your config:

{
  "channels": {
    "whatsapp": {
      "allowFrom": ["+1234567890"]
    }
  }
}

Replace the phone number with your own. This is a security measure — without it, anyone who finds your bot's number could chat with your AI assistant (and use your API credits).

[IMAGE: Screenshot of terminal showing the QR code for WhatsApp linking]

[IMAGE: Screenshot of WhatsApp conversation with the AI assistant responding to a first message]

Step 5: Start the Gateway

The gateway is OpenClaw's control center. It manages all your channels, sessions, and tools from one place.

If you used --install-daemon during onboarding, the gateway is already running in the background. You can check its status with:

openclaw gateway --port 18789 --verbose

The --verbose flag shows detailed logs so you can see exactly what's happening — helpful when you're first getting started.

You can also open the control dashboard in your browser at http://localhost:18789.

[IMAGE: Screenshot of the OpenClaw control dashboard in a browser, showing the main status panel]

Step 6: Send Your First Message

Everything is connected. Time to talk to your assistant.

Open WhatsApp or Telegram and send a message like:

"Remind me to call the dentist tomorrow at 9am"

Or:

"What's on my calendar this week?"

Or simply:

"Hey, what can you help me with?"

Your assistant should respond within a few seconds. If it does — congratulations, you've got your own personal AI assistant running.

[IMAGE: Screenshot showing a natural conversation with the assistant — a reminder being set, then confirmed]

What Can Your Assistant Actually Do?

Now that it's running, here's a quick overview of what OpenClaw can handle out of the box:

Email management — Connect your Gmail and your assistant can read, draft, and organize emails. It can summarize your inbox, flag important messages, and draft replies for you to approve.

Calendar and scheduling — It reads your calendar, creates events, and sends you reminders through WhatsApp or Telegram at the times you choose.

Task tracking — Tell it to remember things, create to-do lists, and follow up with you. It maintains context across conversations, so it remembers what you asked for last week.

Web browsing — It can search the web, read articles, and summarize information for you. Need to research something? Just ask.

Smart home control — If you're running Home Assistant or similar, OpenClaw can control your lights, thermostat, and other devices through natural conversation.

Automation — Set up cron jobs, webhooks, and triggers. "Every Monday at 8am, summarize my unread emails and send me the summary on WhatsApp" — that kind of thing.

Common Issues and How to Fix Them

"The bot isn't responding"

First, check that the gateway is running:

openclaw doctor

This diagnostic tool checks your entire setup and flags anything that's misconfigured. It's the fastest way to find problems.

"I get an API error"

Make sure your Anthropic or OpenAI API key is valid and has credits. You can check your balance at console.anthropic.com or platform.openai.com.

"WhatsApp disconnected"

WhatsApp links expire if your computer goes offline for too long. Re-link by running:

openclaw channels login

And scanning the QR code again.

"Messages are slow"

If responses take more than 10 seconds, try switching to a faster model. Edit your config:

{
  "agent": {
    "model": "anthropic/claude-sonnet-4-5"
  }
}

Sonnet is faster and cheaper than Opus, with only a small trade-off in capability.

Running OpenClaw 24/7 (Optional but Recommended)

If you close your laptop, your assistant goes offline. For an always-on setup, you have a few options.

Option 1: Keep Your Computer Running

The simplest approach. If you have a desktop or Mac Mini that stays on, this works fine.

Option 2: Use a Cloud Server

A small Linux VPS gives you a dedicated always-on assistant. DigitalOcean, Hetzner, and others offer servers for $5–12/month that work perfectly for this.

DigitalOcean offers a 1-click OpenClaw deployment in their marketplace, which is the easiest cloud setup:

  1. Sign up at DigitalOcean (new accounts get $200 in free credits)
  2. Search for "OpenClaw" in the marketplace
  3. Click deploy
  4. SSH into your server and run openclaw onboard

Option 3: Use Docker

If you're comfortable with Docker:

docker-compose up -d

The included docker-compose.yml file handles everything. Your assistant runs in a container that restarts automatically.

[IMAGE: Simple diagram showing the three deployment options — local computer, cloud server, Docker — with pros/cons for each]

Keeping Your Setup Secure

Since OpenClaw connects to real messaging platforms and has access to tools on your machine, security matters. A few important things to configure:

Restrict who can message your bot. Always set allowFrom on every channel so random people can't interact with your assistant.

Use pairing mode for new contacts. By default, OpenClaw uses DM pairing — unknown senders get a pairing code instead of a response. Keep this on.

Run the security check:

openclaw doctor

This flags any risky configurations, like open DM policies or exposed services.

If you're running in the cloud, use Tailscale for secure access to your dashboard instead of exposing it to the public internet.

Updating OpenClaw

The project moves fast. To grab the latest version:

npm install -g openclaw@latest
openclaw doctor

The doctor command checks if your config needs any updates after upgrading.

Next Steps

You've got a working AI assistant. Here are some things to explore next:

Add more channels — Connect Slack, Discord, or Signal so your assistant is available everywhere. Each channel has its own setup guide in the official docs.

Install skills — Skills extend what your assistant can do. Browse available skills at ClawHub or in the control dashboard.

Customize the personality — Edit the SOUL.md file in your workspace (~/.openclaw/workspace/) to change how your assistant talks and behaves.

Set up browser control — Enable the browser tool so your assistant can navigate websites, fill out forms, and extract information for you.

Connect your email — Set up Gmail Pub/Sub triggers so your assistant can proactively notify you about important emails.

Have questions? Something not working? Drop a comment below or join the community to get help from other OpenClaw users.