Install and run OpenClaw no-code setup [with video]

Author: Stef Van Looveren
Topic: OpenClaw installation, Railway deployment, AI agent setup
Difficulty: Beginner to intermediate
Time required: ~15 minutes

This tutorial shows how to install OpenClaw using Railway without writing code. By the end, you’ll have a working AI agent running in the cloud with a connected large language model.

We’ll focus on safety, simplicity, and getting a working AI agent online quickly.

What is OpenClaw?

OpenClaw is an agent-based AI system that can perform tasks, automate workflows, and interact with tools and services through integrations (“skills”). Instead of running only as a chatbot, it behaves more like a programmable assistant.

Because OpenClaw can access files, APIs, and integrations, it’s safer to run it on a separate environment rather than your personal computer.

This tutorial uses Railway as the hosting platform.

Why run OpenClaw in the cloud?

• Security: Avoid exposing personal files or accounts
• Reliability: The agent runs 24/7
• Automation support: Scheduled jobs and integrations
• Easy deployment: Templates install everything automatically

Think of Railway as a “one-click server setup” platform for AI tools.

Go to Railway

Prerequisites

Before starting, make sure you have:

• A GitHub account
• A Railway account
• A credit card added to Railway billing
• An OpenAI API key (or another LLM provider)

Typical hosting cost is about $5/month, depending on usage (use https://railway.com?referralCode=stef to get 20$ free credits!).

Step 1: Sign in to Railway with GitHub

Go to Railway and log in using GitHub.

If you don’t have GitHub yet, create an account:

https://github.com

Using a separate account for AI agents is recommended.

Step 2: Create a new Railway project

Inside Railway:

• Click New Project
• Select Template
• Search for the OpenClaw Railway template

The template installs OpenClaw automatically with all required services.

This avoids manual setup, repository cloning, and configuration work.

Step 3: Configure environment variables

The template requires two values:

• Setup password
• OpenClaw gateway token

These credentials protect your OpenClaw instance because it will run publicly on the internet.

After entering them, click Deploy.

Deployment takes a few minutes.

Step 4: Open the OpenClaw setup page

Once deployment finishes:

• Open your Railway project
• Go to Settings → Public Networking
• Open the generated URL

Log in using:

• Gateway token (username)
• Setup password

This opens the OpenClaw setup interface.

Step 5: Connect a Large Language Model

OpenClaw requires a language model provider to function.

Supported providers include:

• OpenAI
• Anthropic
• Google Gemini

In this tutorial, we use OpenAI.

Create an API key here:

https://platform.openai.com/api-keys

Then paste the key into the OpenClaw setup screen and run onboarding.

This connects your AI agent to its reasoning engine.

Step 6: Open the OpenClaw dashboard

After onboarding completes, open the OpenClaw UI.

You’ll see:

• Agent dashboard
• Sessions
• Usage metrics
• Configuration tools
• Skills integrations

Enter your gateway token again to connect the dashboard.

Understanding how OpenClaw agents work

OpenClaw agents are defined using configuration files such as:

• identity.md
• user.md
• soul.md
• tools.md

These files define:

• Agent personality
• User identity
• Capabilities
• Behavior rules

For example, when you name the agent during setup, OpenClaw writes this information into identity.md.

Sessions, jobs, and automation

OpenClaw supports persistent sessions and scheduled jobs.

Example automation:

• Check tomorrow’s weather every night
• Send a daily email summary
• Run background workflows

These are configured in the Cron Jobs section.

Skills and integrations

Skills extend OpenClaw with external tools.

Examples include:

• Google Workspace integration
• Email automation
• Calendar updates
• Image generation tools
• Hardware integrations

Skills work like plugins for your AI agent.

Testing your OpenClaw installation

Once everything is configured, test your agent by sending a message in the dashboard.

If the agent responds, your installation is working correctly.

You now have a fully deployed AI agent running in the cloud.

Common beginner mistakes

• Running OpenClaw on a personal machine
• Using personal email accounts for agent integrations
• Forgetting to add billing in Railway
• Missing API keys