OpenClaw is an open-source AI agent platform that lets you run autonomous AI assistants on your own hardware. Whether you want full privacy without sending data to OpenAI or Anthropic, or you want to experiment with local LLMs, OpenClaw provides a complete framework for building and deploying AI agents.
What You Need First
Before installing, ensure your machine meets these minimum requirements: a 2-core CPU, 4GB RAM, and 100GB of free disk space. Ubuntu 24.04 LTS is recommended for the best experience, though macOS and Windows (via WSL2) work fine.
You'll also need two core dependencies:
- Node.js v22+ — Download from nodejs.org or use a version manager like nvm
- Git — For pulling updates and managing the installation
Verify both are installed by running node --version and git --version in your terminal.
Installation: The One-Line Method
OpenClaw provides a universal installer that handles OS detection and dependency installation automatically. Run this command in your terminal:
curl -fsSL https://openclaw.ai/install.sh | bash
For Windows, open PowerShell as Administrator and run:
iwr -useb https://openclaw.ai/install.ps1 | iex
The script will detect your operating system, install Node.js if missing, and launch the onboarding process. Installation typically completes in 10-15 minutes on modern hardware.
Configuring Your AI Provider
After installation, the onboarding wizard asks you to choose an AI provider. Three main options exist:
- Anthropic (Claude) — Best for general reasoning, requires an API key from console.anthropic.com
- OpenAI (GPT models) — Good for wide compatibility, get your key from platform.openai.com
- Ollama (fully local) — No API costs, runs models like Qwen 8B or Llama 3 locally on your GPU
To configure with Anthropic, run:
openclaw onboard --anthropic-key YOUR_KEY
For local-only operation with Ollama, select it during onboarding and pull a model with ollama pull qwen:8b.
Starting and Testing
Once configured, start OpenClaw with:
openclaw
The control UI becomes available at http://localhost:18789. From here you can add skills from the ClawHub marketplace, configure Telegram or WhatsApp integrations, and monitor your agent's activity.
Run openclaw doctor to verify everything is working correctly and diagnose any issues.
Why Run Locally?
Privacy is the main driver — your conversations never leave your machine. There's no API bill either once you've invested in GPU hardware, and you maintain full control over your AI assistant without depending on external services.