Getting Started

OpenClaw is a self-hosted personal AI assistant that runs on your own hardware. This guide answers the essentials: what it is, why people choose it over cloud chatbots, and how to get from zero to a working assistant in about 15 minutes.

Quick Start checklist

• □ Install OpenClaw (one-liner, npm, or Setup Wizard; with Ollama: ollama launch openclaw)
• □ Run openclaw onboard — set LLM provider and add a channel
• □ Send your first message via the channel or WebChat

First 24 hours

• If the agent replies but never runs tools (files, commands, browser), set "tools": { "profile": "full" } in config — see FAQ: Agent doesn't run tools and Troubleshooting.
• Run openclaw backup create before changing a lot of config — see FAQ: How to back up and Workspace Backup.
• Bookmark FAQ and Troubleshooting for common first-hour issues.

OpenClaw is an open-source AI assistant platform that runs entirely on your machine. It connects to your preferred LLM (Claude, GPT, local models) and communicates through messaging apps you already use—WhatsApp, Telegram, Discord, Slack, Signal, iMessage, Microsoft Teams, WebChat, and more. Your conversations, memories, and configuration stay on your devices. The only external connection is to your chosen LLM provider for inference.

Unlike a web-based chatbot, OpenClaw gives you a persistent assistant that can automate tasks, use skills (calendar, email, GitHub, smart home), run browser automation, and maintain long-term memory—all under your control.

• Data privacy — Your conversations and memory files live on your machine, not in someone else's cloud.
• No vendor lock-in — Switch LLM providers, add channels, or modify the code. You own the setup.
• Customization — Hundreds of skills on ClawHub; create your own; route different channels to different agents.
• Cost control — Pay for API usage (or use local models with Ollama) instead of fixed subscriptions.
• Full feature set — Browser automation, voice, multi-agent teams, and integrations that cloud chatbots don't offer.

Read more: Why Self-Host AI? →

ChatGPT, Claude.ai, and similar services are great for quick queries. OpenClaw is different:

• Self-hosted — Runs on your hardware; you control data and infrastructure.
• Multi-channel — Use it from WhatsApp, Telegram, Discord, etc.—wherever you chat.
• Persistent memory — Learns your preferences and context over time; editable Markdown files.
• Skills and automation — Calendar, email, code, smart home, scheduled tasks.
• LLM choice — Connect Claude, GPT, Ollama, Gemini, or others; mix and match.

Choose OpenClaw when you want a personal assistant you own and can extend. Choose ChatGPT when you want a quick web chat with no setup.

Follow these steps in order:

Wondering what it costs to run OpenClaw? See our Example Setups & Cost guide for typical configurations: local vs VPS, tiered model routing, and how to keep API costs manageable.