What Is OpenClaw? A Complete Guide to the Open-Source AI Assistant

What Is OpenClaw?

OpenClaw is a free, open-source personal AI assistant that runs on your own devices. Created by Austrian developer Peter Steinberger and originally released in November 2025 under the name Clawdbot, it was briefly renamed Moltbot before settling on OpenClaw in late January 2026.

Unlike cloud-only assistants like ChatGPT or Gemini, OpenClaw runs locally — your files, emails, and personal data never leave your machine unless you tell it to reach out. Because it has direct access to your operating system, it can browse the web, read and write files, run shell commands, and take autonomous actions on your behalf.

The project exploded in popularity in January 2026, jumping from 9,000 to over 60,000 GitHub stars in a matter of days, making it one of the fastest-growing open-source projects in history.

How OpenClaw Works

At its core, OpenClaw is a local gateway — a WebSocket-based control plane that coordinates sessions, messaging channels, tools, and events. It runs as a daemon on your machine and connects to the AI providers you choose.

Your Channels (WhatsApp, Slack, Discord...)
        ↓
   OpenClaw Gateway (ws://127.0.0.1:18789)
        ↓
   LLM Provider (Anthropic, OpenAI)
        ↓
   Skills & Tools (files, browser, APIs...)

You interact with your assistant through the messaging platforms you already use. OpenClaw handles the orchestration between your messages, the AI model, and whatever tools or skills are needed to complete the task.

Key Features

12+ Messaging Channels

OpenClaw supports a wide range of channels out of the box:

• WhatsApp, Telegram, Signal — personal messaging
• Slack, Microsoft Teams, Google Chat — workplace tools
• Discord — communities and gaming
• iMessage (via BlueBubbles), Matrix, Zalo — additional options
• WebChat — built-in browser-based interface

You can connect multiple channels simultaneously, and even route different channels to isolated agent sessions using multi-agent routing.

Bring Your Own LLM

OpenClaw isn't locked to a single AI provider. It supports:

• Anthropic (Claude) — recommended for its long-context strength
• OpenAI (GPT-4o, o1, etc.) — widely available with broad tool support

You authenticate with your own API key or OAuth credentials. You can configure model selection and failover, so if one provider has an outage, your assistant switches to a backup automatically.

Skills and ClawHub

OpenClaw uses a plugin system called Skills — modular extensions that teach the agent how to use specific tools. Each skill is a directory containing a SKILL.md file with YAML frontmatter and natural-language instructions.

The public skill registry, ClawHub, hosts over 5,700 community-built skills covering:

• Productivity (Google Calendar, Todoist, Notion)
• Development (GitHub, CI/CD, code review)
• Smart home (Home Assistant, IoT devices)
• Communication (email drafting, auto-replies)
• Research (web scraping, data synthesis)

Installing a skill is a single command:

openclaw skill install google-calendar

All skills uploaded to ClawHub are scanned with VirusTotal threat intelligence — skills flagged as suspicious are marked with warnings, while verified skills are auto-approved.

Voice and Visual Capabilities

OpenClaw includes built-in voice support powered by ElevenLabs:

• Voice Wake — always-on speech activation
• Talk Mode — natural conversation with voice input and output

Live Canvas provides an agent-driven visual workspace for tasks that benefit from a graphical interface, like diagram generation or data visualization.

Browser Control

OpenClaw can launch and control a dedicated Chrome/Chromium instance. This lets it fill out forms, take screenshots, extract data from web pages, and navigate sites autonomously — useful for tasks like booking appointments, checking prices, or monitoring dashboards.

Persistent Memory

Unlike stateless chatbots, OpenClaw maintains context across conversations. It remembers your preferences, past discussions, and accumulated knowledge. This makes it genuinely useful as a long-running assistant that improves over time.

Supported Platforms

OpenClaw runs across the major operating systems:

The gateway itself requires Node.js 22+ and can be installed via npm, pnpm, or bun.

Installation

Getting OpenClaw running takes a few minutes:

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

# Windows (PowerShell)
iwr -useb https://openclaw.ai/install.ps1 | iex

# Or via npm
npm install -g openclaw@latest

Then run the onboarding wizard:

openclaw onboard --install-daemon

This walks you through gateway setup, channel configuration, and skill installation. Once complete, access the dashboard at http://127.0.0.1:18789/.

Self-Hosting vs Managed Hosting

Running OpenClaw yourself means managing a server 24/7 — security patches, Docker configuration, SSL certificates, backups, and uptime monitoring. Minimum requirements for a VPS are 2 CPU cores, 2 GB RAM, and 20 GB SSD storage.

For users who want OpenClaw running around the clock without the operational overhead, managed hosting handles all of that. ClawNest provides fully managed OpenClaw instances with automatic backups, a web dashboard, EU data residency, and deployment in minutes rather than hours.

Read our detailed comparison: Managed Hosting vs Self-Hosting.

Who Uses OpenClaw?

OpenClaw powers a wide range of use cases:

• Personal assistants — daily planning, email triage, knowledge management
• Customer support — first-line inquiries, ticket routing, FAQ handling
• Development — code review, documentation generation, CI/CD automation
• Research — web monitoring, data synthesis, trend analysis
• Community management — Discord AI assistants, onboarding flows, moderation

See our full list of use cases for more examples.

Getting Started

If you want to try OpenClaw without the complexity of self-hosting, ClawNest offers a free 3-day trial — no credit card required. Follow our step-by-step tutorial to deploy your first AI assistant in under 5 minutes.