OpenClaw Quickstart Guide

quickstartbeginner4 min readVerified Mar 8, 2026

Go from zero to a running OpenClaw AI agent in under 30 minutes with this step-by-step quickstart guide.

openclawquickstartgetting-startedinstallation

OpenClaw Quickstart Guide

This section walks you through everything you need to get a working OpenClaw agent — from installation to your first automated workflow. Most builders can complete the full quickstart in under 30 minutes.

Before You Start#

You will need:

  • A computer — macOS, Linux, or Windows (via WSL2). Start on your laptop; you can move to a VPS later.
  • Node.js 22+ — OpenClaw runs as a long-running Node.js process.
  • An LLM API key — Anthropic (Claude), OpenAI, or a local model via Ollama.
  • A messaging account — Telegram is the easiest starting point.
Tip

Start on your laptop. You do not need a Mac Mini or a remote server to get going. Install OpenClaw on the computer you already use, and move to a dedicated device later if you want the agent running 24/7.

## The Setup Path

OpenClaw setup has four stages, each covered in its own tutorial:

1. Installation#

Install OpenClaw via npm and run the onboarding wizard. Takes about 5 minutes.

2. Messaging Setup#

Connect your first messaging platform — Telegram is recommended for beginners. Takes about 5 minutes.

3. LLM Setup#

Configure your AI model — Claude, GPT, or a local model via Ollama. Takes about 5 minutes.

4. First Agent#

Build your first real workflow: a morning briefing that summarizes your email, calendar, and news. Takes about 15 minutes.

Architecture Overview#

Understanding how OpenClaw works helps you troubleshoot and customize:

[You] --> [Messaging App] --> [Gateway] --> [LLM Provider] --> [Skills/Tools]
  ^                             |
  +---------- Response ---------+
  • Gateway: A long-running Node.js process that listens for messages from all connected platforms.
  • LLM Provider: The AI brain — Claude, GPT, Ollama, or others.
  • Skills: MCP-based plugins that give your agent abilities (file access, web browsing, email, etc.).
  • Memory: Persistent context stored in ~/.openclaw/ that survives restarts.

The Gateway is always running. When you send a message on Telegram, it catches it, forwards it to the LLM with context, and delivers the response back.

Choosing Your Setup Path#

| Path | Best For | Cost | Uptime | |------|----------|------|--------| | Laptop | Getting started, experimenting | Free + API costs | Only when laptop is on | | VPS (Docker) | Always-on agent | $5-20/month + API costs | 24/7 | | Dedicated Mac Mini | Power users, multi-agent | $500+ upfront + API costs | 24/7 |

For most builders, starting on a laptop and moving to a $5/month VPS when ready is the sweet spot.

Warning

Before exposing OpenClaw to the internet (VPS deployment), read the Security Guide. Default settings are not safe for production use.

PreviousOpenClaw vs n8nNextMessaging Platform Setup