Running your own AI assistant locally is one of the best ways to keep control of your data while experimenting with powerful autonomous agents.
In this guide, you’ll learn how to install OpenClaw (formerly known as Clawdbot → Moltbot) on a local Mac Mini, connect it to messaging platforms, and run it as a background service.
By the end, you’ll have a self-hosted AI agent responding to messages and powered by your own API keys — all on your own hardware.
Important 2026 update
The project launched as Clawdbot (Nov 2025), briefly became Moltbot, and is now officially OpenClaw (since late January 2026).
The CLI command has transitioned toopenclaw, but many older tutorials still showclawdbotormoltbot. This guide uses the current commands (February 2026).
What Is OpenClaw?
OpenClaw is an open-source, self-hosted AI assistant designed to run locally on your machine.
Unlike cloud bots, OpenClaw:
- Runs entirely on your Mac
- Connects to WhatsApp, Telegram, Slack, Discord, iMessage and more
- Uses your own API keys (OpenAI, Google Gemini, local models via Ollama)
- Can execute real workflows, control apps, browse, send emails, manage files & calendars
- Acts as a persistent, proactive personal AI agent
Official site: https://openclaw.ai
Docs: https://docs.openclaw.ai
Prerequisites
- A Mac Mini (Intel or Apple Silicon — M-series models work especially well)
- macOS Ventura (13) or newer recommended (Sonoma / Sequoia ideal)
- Internet access
- Basic Terminal knowledge
- An API key from a language model provider
- Node.js 22 or newer
Step 1 — Open Terminal
Press:
Command + SpaceSearch for Terminal, then open it.
yourname@MacMini ~ %All commands below go here.
Step 2 — Install Homebrew (Recommended)
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"Verify:
brew --versionStep 3 — Install Node.js
OpenClaw requires Node 22+.
brew install nodeConfirm:
node --versionv22.x.xStep 4 — Install OpenClaw
Option A — Official Install Script (Recommended)
curl -fsSL https://openclaw.ai/install.sh | bashOption B — npm Install
npm install -g openclaw@latestStep 5 — Run the Onboarding Wizard
openclaw onboard --install-daemon| Onboarding step | What you’ll do |
|---|---|
| 1) Local installation mode | Choose the local/self-hosted install option. |
| 2) Enter API key | Add the provider key you want OpenClaw to use. |
| 3) Choose default model | Select the model OpenClaw should use by default. |
| 4) Select messaging platforms | Pick which channels to connect (WhatsApp, Telegram, etc.). |
| 5) Install background daemon (launchd) | Install the macOS service so it stays running after you close Terminal. |
The daemon keeps OpenClaw running even when Terminal is closed.
Step 6 — Connect Messaging Platforms
- WhatsApp (QR)
- Telegram (bot token)
- Slack
- Discord
- iMessage
Step 7 — Start / Verify the Gateway
openclaw gateway --port 18789 --verboseStep 8 — Test Your Bot
Hello — what can you do?If everything is working, OpenClaw replies instantly.
Security Best Practices
- Never install untrusted skills
- Protect API keys
- Run under a dedicated macOS user
- Do NOT expose port 18789 publicly
- Review permissions carefully
npm update -g openclawOptional: Uninstalling OpenClaw
openclaw uninstallFinal Thoughts
If setting up and maintaining a local AI agent feels like more overhead than you need, ClawPilot offers a cloud-hosted alternative that gives you AI agent capabilities without the local installation, daemon management, or hardware requirements.
You now have OpenClaw running locally on your Mac Mini — a fully self-hosted, always-on AI assistant connected to real messaging apps and powered by the models you choose.
- Add custom skills & workflows
- Integrate APIs & calendars
- Run local LLMs (Ollama / LM Studio)
- Build personal automation agents
All without depending on third-party cloud services.