How to Set Up OpenClaw in 2026: Complete Step-by-Step Guide

Most people use AI through chat interfaces -- ChatGPT, Claude, Gemini. You type a question, you get an answer. But an AI agent is different. An agent does not just answer questions; it takes actions on your computer. It clicks buttons, fills forms, reads files, sends emails, and executes multi-step workflows autonomously.

The demand for this is growing fast. A 2025 Gartner forecast projected that by 2028, 33% of enterprise software applications will include agentic AI. And developers are not waiting -- tools like OpenClaw have attracted hundreds of thousands of GitHub stars because they let anyone build and run their own AI agent without relying on a paid SaaS platform.

There are three main reasons people set up their own agents:

Cost control. Managed AI agent platforms charge monthly subscriptions. With an open-source tool like OpenClaw, you only pay for the underlying model API calls (typically $5-20/month for moderate usage with Anthropic or OpenAI).

Customizability. OpenClaw uses a plugin and skill system that lets you build custom automations. You can modify the agent's behavior, add new capabilities, and integrate with any tool that has an API.

Data privacy. When you self-host, your prompts and data stay on your machine. No third-party platform sees your queries. For teams handling sensitive data, this is a hard requirement.

OpenClaw is an open-source AI agent framework that turns large language models into autonomous computer-use agents. It connects to model providers like Anthropic (Claude), OpenAI (GPT), and Google (Gemini) through their APIs, then extends those models with the ability to take actions -- browsing the web, executing shell commands, reading and writing files, and controlling desktop applications.

Key technical facts (source):

• License: MIT (fully open source, free to use commercially)
• Runtime: Node.js 22.14+ (Node 24 recommended)
• GitHub stars: 361K+ (as of April 2026)
• Architecture: Gateway service running on port 18789 + CLI or dashboard UI for interaction
• Model support: Anthropic Claude, OpenAI GPT, Google Gemini, and local models via Ollama
• Messaging integrations: Telegram, Discord, WhatsApp, and Slack channels (source)

OpenClaw is not a model itself -- it is the framework that sits between you and the model. You bring your own API key, and OpenClaw provides the scaffolding that turns a language model into an agent that can actually do things on your computer.

OpenClaw is the right choice if you meet all three of these criteria:

1. You are comfortable with a terminal. The setup process involves running CLI commands, editing JSON configuration files, and potentially managing a background service. If you have never used a terminal, expect a learning curve.
2. You want maximum control over your agent. OpenClaw gives you access to the full source code, plugin system, and configuration. You can modify anything. If you just want an agent that works out of the box, a managed platform is a better fit.
3. You understand the security tradeoffs. OpenClaw runs with full local access to your machine. There is no built-in approval system for dangerous actions -- the agent can read, write, and delete any file your user account can access. The security model relies on application-level checks within a shared memory process, not OS-level isolation.

Who should NOT use OpenClaw:

• Non-technical users who do not want to manage prerequisites, configuration files, or server deployments
• Anyone handling sensitive data who needs OS-level isolation or a human-in-the-loop approval system before the agent executes destructive actions
• Teams that need 24/7 availability without managing their own infrastructure -- OpenClaw requires you to host and maintain the Gateway service yourself

If any of these describe you, skip to the "How to Use Sai for This" section at the end for a zero-setup alternative.

Step 1: Install Prerequisites

You need Node.js 22.14 or higher (Node 24 recommended) and an API key from a model provider.

Check your Node.js version:

node --version

If you do not have Node.js or your version is too old, install it from nodejs.org or use a version manager:

nvm install 24
nvm use 24

Get an API key:

OpenClaw supports Anthropic (Claude), OpenAI (GPT), Google (Gemini), and local models via Ollama. For Anthropic, sign up at console.anthropic.com. For OpenAI, visit platform.openai.com.

‍

Step 2: Install OpenClaw

Run the appropriate command for your OS (source):

Mac / Linux:

curl -fsSL https://openclaw.ai/install.sh | bash

Windows (PowerShell):

irm https://openclaw.ai/install.ps1 | iex

Alternative methods: Docker, Nix (nix-openclaw), or npm (npm install -g openclaw).

Verify installation:

openclaw --version

‍

Step 3: Run the Onboarding Wizard

openclaw onboard --install-daemon

The wizard asks you to (source):

1. Choose a model provider -- Anthropic (Claude), OpenAI, Google, or local via Ollama
2. Enter your API key -- stored in ~/.openclaw/openclaw.json
3. Configure the Gateway -- the background service that handles agent communication on port 18789

The --install-daemon flag sets up the Gateway as a system service so it runs in the background.

‍

Step 4: Verify and Start Using Your Agent

Confirm the Gateway is running:

openclaw gateway status

Then start chatting with your agent:

openclaw              # Terminal chat (quickest)
openclaw dashboard    # Visual UI in your browser

‍

Step 5: Connect Messaging Channels (Optional)

OpenClaw supports Telegram, Discord, WhatsApp, and Slack. Each channel requires its own bot token.

Telegram example:

1. Open Telegram, search for @BotFather
2. Send /newbot and follow the prompts
3. Copy the bot token to ~/.openclaw/openclaw.json
4. Restart: openclaw gateway restart

‍

Step 6: Deploy for 24/7 Operation (Optional)

To keep your agent always available, run the Gateway on a VPS (DigitalOcean, Hetzner, or AWS Lightsail -- $5-10/month for a basic server).

openclaw gateway status    # Check if running
openclaw gateway start     # Start the service
openclaw gateway stop      # Stop the service
openclaw gateway restart   # Restart after config changes

On your personal machine, the Gateway stops when you shut down your computer.

‍

OpenClaw Setup vs. Alternatives: How Do They Compare?

‍

How to Use Sai for This

If you want an AI agent that controls your desktop and browser without any terminal setup, server management, or API key configuration, Sai by Simular is a managed alternative that works out of the box.

• Zero setup required -- download the desktop app, sign in, and start automating tasks in under 2 minutes. No Node.js, no API keys, no configuration files (source)
• Built-in security -- every potentially destructive action (sending emails, deleting files, submitting forms) requires your explicit approval before Sai executes it, with cloud VM isolation separating the agent from your local machine (source)
• Desktop + browser + integrations included -- Sai controls native apps and browser tabs, connects to Gmail, Google Calendar, Google Sheets, and GitHub natively, and supports scheduled workflows -- all for 7 days free trail with no per-API-call charges

‍