Install a complete AI agent in 10 minutes. Zero technical knowledge required.
npx openclaw
Latest version, all dependencies, properly configured. No manual steps.
Telegram, Discord, Slack, WhatsApp — pick one and it's set up for you.
Claude, GPT, Gemini, DeepSeek, or local Ollama — your choice, ready to go.
Choose from 5 pre-built SOUL.md templates or create your own during setup.
Firewall, SSH hardening, auto-updates — production-ready from day one.
Systemd service configured. Your agent survives reboots automatically.
We benchmarked OpenClaw EasySetup against other installation methods to show you exactly what you're getting.
| Method | Setup Time | Technical Skill | Success Rate |
|---|---|---|---|
| OpenClaw EasySetup | 10-15 minutes | Beginner | 98% |
| Manual Installation | 2-4 hours | Advanced | ~70% |
| Docker Compose | 30-45 minutes | Intermediate | ~85% |
| Managed Hosting | 5 minutes | Beginner | 99% |
| Configuration | RAM Usage | CPU Load | Disk Space |
|---|---|---|---|
| Single agent, 1 channel | ~200MB | <5% | ~500MB |
| Single agent, 3 channels | ~400MB | <8% | ~700MB |
| 3 agents, multi-channel | ~1.2GB | <15% | ~1.5GB |
| Heavy use + browser skills | ~2.5GB | <30% | ~3GB |
Key takeaway: A basic $5/month VPS with 1GB RAM handles a single OpenClaw agent easily. For multiple agents or browser automation, upgrade to 4GB RAM ($5-7/month).
Buy a VPS from any provider ($5-12/mo). We recommend Hostinger for beginners.
Open Terminal (Mac) or PuTTY (Windows) and connect to your server. Your VPS provider gives you the IP and password.
npx openclaw
The script asks simple questions (your API key, which channel, etc.) and does everything else.
Open your messaging app and send a message to your new AI agent. That's it — you're done!
If EasySetup doesn't work on your server, we'll refund you. No questions asked.
No. If you can copy-paste a command, you can use EasySetup. The script asks simple questions and handles everything else.
Any VPS running Ubuntu 22.04+, Debian 12+, or similar Linux. Tested on Hostinger, DigitalOcean, Vultr, Hetzner, and Oracle Cloud.
Claude (Anthropic), GPT-4o (OpenAI), Gemini (Google), DeepSeek, local models via Ollama, and OpenRouter (all models).
Telegram, Discord, Slack, WhatsApp, Signal, and more. You choose during setup.
Yes! You can read the full source code before running it. We believe in transparency.
The script detects existing installations and offers to update/reconfigure instead of reinstalling.
Yes, the AI model is a separate cost ($0-50/month depending on usage). See our cost guide.