OpenClaw on a Raspberry Pi

Everyone's out here buying Mac Minis or repurposing old laptops to run OpenClaw. But there's a better option hiding in plain sight: a Raspberry Pi. It's pocket-sized, consumes barely any power, and costs a fraction of any alternative.

The question was: can it actually run OpenClaw? The answer — yes. And this guide shows you exactly how to set it up from scratch.

Why Raspberry Pi Is the Most Underrated AI Agent Hardware

The Raspberry Pi might be tiny, but it punches well above its weight for running an always-on AI agent:

• Under $100 — hardware + accessories, all in
• Always on — uses ~5W of power (your electricity bill won't notice)
• No subscription — it's your hardware, running your agent
• Silent and compact — fits anywhere, runs anywhere
• Genuinely capable — ARM64 with enough RAM to run OpenClaw comfortably

What You Need Before Installing OpenClaw on Raspberry Pi

Step 1: Set Up Raspberry Pi OS (64-Bit is Non-Negotiable)

If your Pi came in a kit, your micro SD card may already have Raspberry Pi OS loaded. Make sure it's the 64-bit version — this is the single most important requirement. OpenClaw will not work on the 32-bit (ARMv7) image.

Flash the 64-bit image to your SD card using Raspberry Pi Imager. Insert the card, assemble your Pi, connect your monitor and keyboard, and boot it up. Go through the initial setup prompts on screen.

Step 2: Update Your System and Install the Basics

Once you're at the desktop, open your terminal. The first thing to do is bring everything up to date and install the tools we'll need.

Update your package list:

sudo apt update

Run the full upgrade (this takes a few minutes on a fresh install):

sudo apt full-upgrade -y

Once complete, reboot:

sudo reboot

After your Pi comes back up, open terminal again and install the essential build tools:

sudo apt install git curl ca-certificates build-essential

This takes about a minute. Git, curl, and build-essential are all needed for what comes next.

Step 3: Verify Your Architecture (64-Bit Check)

Before installing anything else, do a quick sanity check to confirm you're running 64-bit:

uname -m

You want the output to end in 64 — specifically aarch64. If you see armv7l, stop here and re-flash your SD card with the 64-bit Raspberry Pi OS image before continuing.

Step 4: Install Node.js on Raspberry Pi (Version 22+)

OpenClaw requires Node.js — and Raspberry Pi doesn't come with it. You'll need version 22 or higher. The easiest way is via NodeSource:

curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash -
sudo apt install -y nodejs

Once installation finishes, verify the versions:

node -v
npm -v

You want to see v22.x.x for Node and a matching npm version. If you see that, you're good to go.

Step 5: Fix npm Permissions Before You Install OpenClaw

This step is crucial and often skipped. By default, npm global installs require sudo on Raspberry Pi — that's a security problem. Here's the recommended fix: set up a custom global directory and add it to your PATH.

Create the npm global directory:

mkdir ~/.npm-global

Point npm to use it:

npm config set prefix ~/.npm-global

Add it to your PATH and reload your shell:

echo 'export PATH=~/.npm-global/bin:$PATH' >> ~/.bashrc
source ~/.bashrc

Confirm npm is accessible from the new path:

which npm

Step 6: Install and Onboard OpenClaw

This is the moment of truth. With Node.js and npm permissions sorted, install OpenClaw globally:

npm install -g openclaw

Once it's done (takes a minute or two), run the onboarding wizard:

openclaw onboard

The onboarding wizard will walk you through:

1. Acknowledge the power + risk — hit yes, you understand what you're enabling
2. Quick start — choose the quick setup path
3. Choose your AI model — OpenAI, Anthropic Claude, or any other supported provider. OpenAI is a good starting point.
4. Choose your channel — pick WhatsApp so you can message your bot from your phone
5. Follow the channel setup instructions — skip anything you want to add later

Connect Your AI Agent to WhatsApp (Remote Access)

The real magic is what happens next. With OpenClaw connected to WhatsApp, you can message and control your Pi-hosted AI agent from anywhere — without being physically near the device.

This means your Raspberry Pi can sit quietly in a corner of your home (or office, or garage), always on, always listening — and you can ping it from your phone, get tasks done, and monitor what's happening remotely via WhatsApp.

This is how you end up with something like "Rocky Pi" — a fully autonomous AI assistant running on hardware that fits in your pocket, that you can message from anywhere in the world.

Is OpenClaw on Raspberry Pi Worth It? Final Verdict

Absolutely. Here's the breakdown:

• Hardware cost: ~$70–100 for Pi + accessories
• Running cost: ~$2–5/month in electricity
• Cloud cost: $0 — it's your hardware
• Performance: More than enough for a personal AI agent
• Setup time: ~30–45 minutes following this guide

If you've been putting off setting up a personal AI agent because of the cost or complexity — a Raspberry Pi removes both barriers. It's genuinely the most accessible way to run OpenClaw.

Frequently Asked Questions