BestSuperintelligent
Explore evaluations

Guide

What Is OpenClaw? The Self-Hosted AI Assistant Explained (2026)

Self-hosted agent guide.

OpenClaw is a free, open-source AI assistant you run on your own hardware. It lives in WhatsApp, Telegram, Slack, Discord and 20+ other apps, supports any AI model, and keeps your data local.

By Liyam Flexer, Founder & EditorPublished Last reviewed
At a glance
Type
Guide
Topics
OpenClaw, AI assistant, Self-hosted AI
Evidence
Editorial interpretation
Written by
Liyam Flexer

The short answer

OpenClaw is a free, MIT-licensed open-source AI assistant that runs on your own hardware and operates directly inside messaging apps like WhatsApp, Telegram, Slack, and Discord. A local Gateway process manages session state, tools, and memory, keeping conversation history and credentials on your machine while allowing swappable frontier and local model backends.

OpenClaw is an open-source, MIT-licensed AI assistant that runs on hardware you control and meets you inside the messaging apps you already use. A single Gateway process on your laptop, Mac mini, Linux machine, or VPS becomes the control plane for sessions, memory, tools, and channels. State, credentials, and conversation history stay on your machine by default. Models and agent harnesses are swappable plugins—you can use Claude, Codex, OpenAI, Gemini, DeepSeek, local models via Ollama, or any other provider without rewriting the rest of the system.

As of early November 2026, the project has approximately 394,000 GitHub stars and is stewarded by the independent OpenClaw Foundation, a 501(c)(3) nonprofit. There is no paid tier, no official hosted service, and no token. You pay only for the models you use and the machine that keeps the Gateway running.

What OpenClaw Actually Does

You interact with OpenClaw the same way you message a person. It supports WhatsApp, Telegram, Discord, Slack, Microsoft Teams, Signal, iMessage, and more than 20 additional channels. Native apps are available for macOS, iOS, Android, Windows, and Linux.

  • Read and write files in its workspace
  • Run shell commands and system scripts
  • Control a browser to search, fill forms, and inspect live pages
  • Search the web and extract source findings
  • Maintain persistent memory across days in plain files you can open and edit
  • Execute scheduled jobs and event-driven automations
  • Use or write custom skills and external tools
  • Operate as a personal assistant or as a shared team Gateway

Sandboxing is available but turned off by default. The architecture follows a “trusted Gateway, untrusted execution, policy enforced by code” model.

Who It Is For

OpenClaw fits people who want an always-available agent inside the chat apps they already open, full control over which model is used, and data that never leaves infrastructure they manage. It is especially popular with developers and power users comfortable maintaining a process and reviewing tool permissions.

Evidence in this article Editorial interpretation Our reading of what the evidence means. Judgment, clearly separated from the evidence itself.

Frequently asked questions

What is OpenClaw?

OpenClaw is an open-source, self-hosted AI assistant that runs as a Gateway process on your hardware and connects directly to 20+ chat apps like Telegram, WhatsApp, and Discord.

Is OpenClaw free to use?

Yes. The OpenClaw software is free and MIT-licensed. You only pay for your underlying model API calls (e.g. Anthropic, OpenAI) or run free local models via Ollama.

Does OpenClaw run locally or in the cloud?

OpenClaw runs its control plane and Gateway on your local machine, Mac mini, or private VPS, keeping your files, conversation history, and credentials on your hardware.

Sources

  1. OpenClaw GitHub Repository OpenClaw Foundation, accessed .
  2. OpenClaw Official Site OpenClaw Foundation, accessed .
  3. OpenClaw Documentation OpenClaw Foundation, accessed .