OpenClaw: How to Set Up Your Own Personal AI Assistant (Complete Guide)

OpenClaw is an open-source personal AI assistant that runs on your own devices and connects to the messaging platforms you already use — WhatsApp, Telegram, Slack, Discord, Signal, iMessage, Microsoft Teams, and more. With over 240,000 stars on GitHub and 920+ contributors, it's one of the most popular open-source AI projects available today.

Unlike cloud-only AI assistants, OpenClaw gives you full control. Your data stays local, you pick the AI model, and you decide which channels it listens on. Whether you want a coding helper, a smart home controller, or a personal research assistant across all your messaging apps, OpenClaw delivers.

What is OpenClaw?

At its core, OpenClaw is a local-first Gateway that acts as a control plane for your personal AI assistant. It connects to AI models (OpenAI, Anthropic Claude, and others) and routes conversations across your messaging channels.

Here's the architecture at a glance:

WhatsApp / Telegram / Slack / Discord / Signal / iMessage / Teams / WebChat
               │
               ▼
┌───────────────────────────────┐
│            Gateway            │
│       (control plane)         │
│     ws://127.0.0.1:18789      │
└──────────────┬────────────────┘
               │
               ├─ AI Agent (RPC)
               ├─ CLI (openclaw …)
               ├─ WebChat UI
               ├─ macOS App
               └─ iOS / Android Nodes

Key highlights:

• Multi-channel inbox — one assistant across WhatsApp, Telegram, Slack, Discord, Google Chat, Signal, iMessage, Microsoft Teams, Matrix, WebChat, and more
• Multi-model support — works with OpenAI, Anthropic Claude, and other providers
• Voice support — Voice Wake and Talk Mode on macOS, iOS, and Android
• Live Canvas — an agent-driven visual workspace
• Browser control — let the assistant browse the web for you
• Skills and plugins — extend functionality with bundled, managed, and custom skills
• Companion apps — macOS menu bar app, iOS node, Android node
• Fully open source — MIT licensed, 240k+ GitHub stars

Prerequisites

Before you begin, make sure you have:

• Node.js 22 or later installed (download here)
• A terminal (macOS Terminal, iTerm2, Windows Terminal with WSL2, or any Linux terminal)
• An API key or OAuth subscription from an AI provider (OpenAI, Anthropic, etc.)

OpenClaw runs on macOS, Linux, and Windows (via WSL2).

Step 1: Install OpenClaw

The recommended way to install OpenClaw is via npm:

npm install -g openclaw@latest

Or if you prefer pnpm:

pnpm add -g openclaw@latest

Verify the installation:

openclaw --version

Step 2: Run the Onboarding Wizard

OpenClaw ships with an interactive onboarding wizard that walks you through the entire setup:

openclaw onboard --install-daemon

The wizard will guide you through:

1. Gateway setup — configuring the local control plane
2. AI model selection — choosing your preferred AI provider and model
3. Channel connections — linking your messaging platforms
4. Skills installation — enabling built-in capabilities
5. Daemon installation — keeping OpenClaw running in the background via launchd (macOS) or systemd (Linux)

The

--install-daemon

Step 3: Configure Your AI Model

OpenClaw stores its configuration in

~/.openclaw/openclaw.json

{
  "agent": {
    "model": "anthropic/claude-opus-4-6"
  }
}

You can use any supported model. The project recommends Anthropic Claude Opus 4.6 for its strong long-context performance and prompt-injection resistance.

For model authentication, you can use OAuth subscriptions or API keys. See the Models documentation for details on all supported providers.

Step 4: Start the Gateway

Once configured, start the Gateway:

openclaw gateway --port 18789 --verbose

The Gateway is now running and ready to accept connections from your messaging channels and the CLI.

Step 5: Connect Your First Channel

Here's how to connect the most popular channels:

WhatsApp

Link your WhatsApp account:

openclaw channels login

Then configure who can talk to the assistant in

openclaw.json

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

Telegram

Create a bot via BotFather, then add the token:

{
  "channels": {
    "telegram": {
      "botToken": "YOUR_BOT_TOKEN"
    }
  }
}

Discord

Create a Discord bot in the Developer Portal, then configure:

{
  "channels": {
    "discord": {
      "token": "YOUR_BOT_TOKEN"
    }
  }
}

Slack

Set up a Slack app with Socket Mode enabled:

{
  "channels": {
    "slack": {
      "botToken": "xoxb-...",
      "appToken": "xapp-..."
    }
  }
}

OpenClaw also supports Signal, iMessage (via BlueBubbles), Microsoft Teams, Google Chat, Matrix, and WebChat out of the box.

Things to Try

Once OpenClaw is running, here are some powerful features to explore:

1. Send Messages from the CLI

You don't need a messaging app to talk to your assistant:

openclaw agent --message "What's on my calendar today?" --thinking high

The

--thinking

2. Use Chat Commands

Send these commands in any connected channel:

/status

/new

/reset

/compact

/think high

/verbose on

/usage full

3. Enable Browser Control

Let your assistant browse the web:

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

Now you can ask your assistant to research topics, fill out forms, or take screenshots of web pages.

4. Set Up Cron Jobs and Webhooks

Automate recurring tasks:

• Cron jobs — schedule the assistant to run tasks at specific times (daily summaries, reminders)
• Webhooks — trigger the assistant from external events
• Gmail Pub/Sub — have the assistant process incoming emails

See the Automation docs for setup.

5. Try Voice Wake and Talk Mode

On macOS, iOS, or Android, enable always-on voice:

• Voice Wake — say a wake word and start talking
• Talk Mode — continuous conversation with ElevenLabs voice synthesis

This turns OpenClaw into a hands-free assistant.

6. Explore the Live Canvas

The Canvas is an agent-driven visual workspace where the assistant can render UI, charts, and interactive content. It support A2UI (Agent-to-UI) for real-time visual updates.

7. Use Agent-to-Agent Communication

OpenClaw supports multi-agent sessions:

• sessions_list

  — discover active agents

• sessions_send

  — message another agent session

• sessions_history

  — fetch transcripts from other sessions

This lets you build workflows where specialized agents coordinate work.

8. Install Skills from ClawHub

ClawHub is OpenClaw's skill registry. With it enabled, your assistant can automatically discover and install new capabilities as needed.

Browse available skills at clawhub.com.

9. Run the Gateway Remotely

Host the Gateway on a Linux server and connect to it from anywhere using Tailscale Serve/Funnel or SSH tunnels:

{
  "gateway": {
    "tailscale": {
      "mode": "serve"
    }
  }
}

Device nodes (macOS, iOS, Android) can still run local actions like camera, screen recording, and notifications via

node.invoke

Building from Source

If you want to contribute or customize OpenClaw:

git clone https://github.com/openclaw/openclaw.git
cd openclaw

pnpm install
pnpm ui:build
pnpm build

pnpm openclaw onboard --install-daemon

For development with auto-reload:

pnpm gateway:watch

Docker Setup

Prefer containers? OpenClaw has Docker support built-in:

git clone https://github.com/openclaw/openclaw.git
cd openclaw
docker-compose up

There's also Podman support for rootless containers. See the Docker guide.

Security Considerations

Since OpenClaw connects to real messaging platforms, security matters:

• DM pairing is on by default — unknown senders get a pairing code instead of a response
• Allowlists — explicitly control who can talk to your assistant per channel
• Sandbox mode — run non-main sessions in Docker containers for isolation
• Run

  openclaw doctor

  to check for misconfigurations

Always review the Security guide before exposing your assistant publicly.

Useful Resources

Final Thoughts

OpenClaw stands out because it's truly personal and local-first. You own your data, pick your AI model, and connect the channels that matter to you. The onboarding wizard makes initial setup painless, and the depth of features — from browser control to voice wake to multi-agent coordination — means you can grow into it over time.

Whether you're a developer looking for a programmable AI assistant, a power user who wants AI across all their messaging apps, or someone who just wants a smarter personal assistant that respects privacy, OpenClaw is worth trying.

Get started in under 5 minutes:

npm install -g openclaw@latest
openclaw onboard --install-daemon

That's it. Your personal AI assistant is ready. 🦞