OpenClaw Getting Started Guide: Build Your Personal AI Assistant in 5 Minutes

Why OpenClaw Is Worth Your Attention

If you want an AI assistant that runs on your own machine or server, but still works inside the chat apps you already use every day, OpenClaw is a strong starting point.

OpenClaw is best understood as a self-hosted AI gateway. It connects chat platforms such as WhatsApp, Telegram, Discord, Feishu, WeChat, and iMessage with the AI models you choose. You send a message from a familiar chat app, OpenClaw forwards it to the model, then sends the reply back to the same app.

That gives you a more private, more flexible, and more customizable setup than a single hosted chatbot product.

How OpenClaw Differs From ChatGPT or Claude

Dimension	OpenClaw	ChatGPT / Claude
Data control	You control deployment and message flow	Mostly controlled by the cloud service
Platform support	Can connect multiple chat platforms	Usually limited to the official app or web UI
Model choice	Can switch between providers	Usually tied to one platform
Customization	Plugins, prompts, per-channel config	More limited customization
Deployment	Self-hosted	Vendor-hosted

If privacy, extensibility, or multi-channel support matters to you, OpenClaw is the more flexible option.

---

Part 1: Install OpenClaw in About 5 Minutes

OpenClaw supports the major desktop and server operating systems. The fastest path is the official install script.

System Requirements

• Node.js: 24.x recommended, 22.14+ also works
• RAM: at least 2GB
• Disk: at least 1GB of free space
• Network: required only if you use cloud model APIs

macOS / Linux

Open a terminal and run:

curl -fsSL https://openclaw.ai/install.sh | bash

The installer typically does four things for you:

1. Detects the current operating system
2. Installs Node.js when it is missing
3. Installs the OpenClaw CLI
4. Starts the onboarding flow

Windows

The recommended option is WSL2, which is usually more stable.

WSL2 flow:

1. Install Ubuntu from the Microsoft Store
2. Open the WSL terminal
3. Run the same install script

curl -fsSL https://openclaw.ai/install.sh | bash

If you prefer native Windows, PowerShell also works:

iwr -useb https://openclaw.ai/install.ps1 | iex

Verify the Installation

openclaw --version

If you see a version string such as:

OpenClaw 2026.3.24

the CLI is installed correctly.

---

Part 2: Complete the Initial Onboarding

Run:

openclaw onboard --install-daemon

The setup wizard usually takes only a couple of minutes.

1. Choose an AI Provider

OpenClaw supports many options, including:

• OpenAI
• Anthropic
• Google Gemini
• Azure OpenAI
• Ollama / LocalAI / LM Studio for local models

For a first setup, OpenAI or Anthropic is usually the easiest choice.

2. Generate the Gateway Token

During onboarding, OpenClaw creates a Gateway token that acts as an admin credential. Keep it somewhere safe.

3. Install as a System Service

--install-daemon installs OpenClaw as a background service, which is useful if you want it to auto-start with the machine.

If you prefer manual control, you can skip that option.

4. Check Status

openclaw status

To inspect logs:

openclaw logs

If the logs show a successful startup message, your gateway is ready.

---

Part 3: Connect Your First Chat Platform

OpenClaw works with many chat platforms. For a quick first test, Telegram is usually the easiest.

Telegram Bot Setup

1. Open Telegram and search for @BotFather
2. Send /newbot
3. Enter the bot name and username
4. Copy the returned Bot Token

Enable Telegram in OpenClaw

Edit your config:

openclaw config edit

Add:

{
  plugins: {
    telegram: {
      botToken: "YOUR_BOT_TOKEN"
    }
  }
}

Then restart:

openclaw restart

Go back to Telegram and send a test message such as:

hello

If you receive an AI reply, the connection is working.

Other Common Platforms

Discord

1. Open Discord Developer Portal
2. Create an application
3. Add a bot and copy the token
4. Put token and clientId into the OpenClaw config

Feishu

Install the plugin first:

openclaw plugins install feishu

Then add appId, appSecret, and any related config values.

WhatsApp / iMessage

These typically require extra bridge tooling such as BlueBubbles, so they are better treated as a second-step setup after the core gateway is already working.

---

Part 4: Send Your First Message

Once a platform is configured, you interact with the assistant exactly where you already chat.

OpenClaw handles the rest:

1. Receives the incoming message
2. Selects the configured model
3. Calls the model
4. Sends the response back to the same chat platform

Example prompts:

Who are you?

Introduce yourself in English

Write a Python quicksort

What's the weather in Beijing?

---

Part 5: Basic Customization

Most of the assistant behavior can be changed through the config file.

Change the Assistant Name and Avatar

openclaw config edit

Example:

{
  agent: {
    name: "XiaoKun",
    avatar: "https://example.com/avatar.png",
    systemPrompt: "You are a helpful AI assistant. Be concise."
  }
}

Set a System Prompt

The system prompt shapes the assistant’s style and behavior.

Example:

{
  agent: {
    systemPrompt: `
You are Kunpeng AI's technical assistant, helping developers with AI and ML questions.

Your style:
- Be concise
- Prefer executable examples
- Say "I'm not sure" when uncertain
- Stay professional on technical questions
    `.trim()
  }
}

Give Each Channel a Different Persona

{
  channels: {
    telegram: {
      agent: {
        name: "Telegram Assistant",
        systemPrompt: "Keep replies short and mobile-friendly."
      }
    },
    discord: {
      agent: {
        name: "Discord Bot",
        systemPrompt: "You can be slightly more playful on Discord."
      }
    }
  }
}

---

Part 6: Useful Commands

openclaw status
openclaw logs
openclaw restart
openclaw config edit
openclaw plugins list
openclaw plugins install <plugin-name>
openclaw plugins uninstall <plugin-name>
openclaw agent --message "your message"
openclaw update
openclaw doctor

---

FAQ

How is OpenClaw different from a normal chatbot?

OpenClaw is more like an AI assistant gateway than a single chatbot. It manages platform integration, model routing, plugins, and automation together.

Can I use local models such as Ollama?

Yes. You can select Ollama during onboarding or configure it manually:

{
  models: {
    default: {
      provider: "ollama",
      model: "qwen2.5:7b",
      baseURL: "http://127.0.0.1:11434"
    }
  }
}

Who is OpenClaw for?

OpenClaw is a great fit if you want to:

• run an AI assistant on your own infrastructure
• connect multiple chat platforms at once
• control prompts, permissions, and plugins
• build a practical AI workflow around the chat tools you already use

Summary

OpenClaw is valuable not because it is “another chatbot,” but because it turns chat platforms, models, plugins, and automation into one system that you control.

If you want an AI assistant that truly belongs to you instead of one locked inside a single hosted app, OpenClaw is a strong place to start.