Setting up your own AI assistant on AWS takes less than 30 minutes and costs under $10/month. This guide walks you through every step — from creating your AWS account to running OpenClaw on a spot instance. No prior cloud experience needed.

Step 1: Sign Up for AWS

The first step is to simply sign up for AWS (Amazon Web Services) if you haven't done so already.

AWS Signup Page

Go to aws.amazon.com and click Create an AWS Account. If you already have an account, click Sign in to an existing AWS account.

Step 2: Search for EC2

Once logged into the AWS Console, search for "EC2" in the search bar at the top.

Search for EC2

Click on EC2 - Virtual Servers in the Cloud to open the EC2 Dashboard.

Step 3: Launch Instance

From the EC2 Dashboard, click the orange Launch instance button.

EC2 Dashboard

This will open the instance launch wizard.

Step 4: Name Your Instance

Give your instance a memorable name. We'll call it "ShaunBot" but you can name it whatever you like.

Name Your Instance

Step 5: Select Ubuntu

Under Application and OS Images, select Ubuntu from Quick Start, then choose Ubuntu Server 24.04 LTS.

Select Ubuntu AMI

Ubuntu 24.04 LTS is free tier eligible and works great with OpenClaw.

Step 6: Select Instance Type

Choose t3.medium as your instance type. This gives you 2 vCPUs and 4 GB of memory - plenty for running OpenClaw.

Select Instance Type

Step 7: Create a Key Pair

You'll need a key pair to connect to your instance via SSH. Click Create new key pair.

Key Pair Section

In the dialog, enter a name for your key pair, select RSA and .pem format, then click Create key pair.

Create Key Pair

Important: Save the downloaded .pem file securely - you'll need it to connect to your instance later.

Step 8: Configure Storage

Scroll down to Configure storage and set it to 20 GiB gp3:

Configure Storage

20GB gives you plenty of room for OpenClaw, logs, and any tools you install.

Step 9: Enable Spot Instance

Under Advanced details, find Purchasing option and select Spot instances. This saves you ~70% on compute costs!

Spot Instance Option

Configure Spot Options

Click Customize Spot instance options to configure the settings:

Spot Instance Options

• Maximum price: Select "No maximum price" (capped at On-Demand price)
• Request type: Choose "Persistent" so it restarts if interrupted
• Interruption behavior: Set to "Stop" to preserve your data

Spot instances are perfect for OpenClaw - they're cheap and any interruptions can be handled gracefully.

Step 10: Review and Launch

Review your configuration in the Summary panel. You should see:

• Ubuntu 24.04 LTS
• t3.large instance
• 8 GiB storage
• Spot instance pricing

Launch Summary

Click Launch instance when ready.

Step 11: Success!

You'll see a success message confirming your instance is launching.

Launch Success

Step 12: View Instance Details

Click on the instance ID to view your instance details. Wait for the Instance state to show Running.

Instance Summary

Note your Public IPv4 address - you'll need this to connect.

Step 13: Connect to Your Instance

Click the Connect button at the top of the instance details page.

EC2 Instance Connect

Select EC2 Instance Connect and click Connect. This opens a browser-based terminal - no SSH setup required!

Step 14: Install Node.js

You're now connected to your Ubuntu server via the browser terminal.

Ubuntu Terminal

First, install Node.js 22.x (required for OpenClaw):

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

Verify the installation:

node --version  # Should show v22.x.x
npm --version   # Should show 10.x.x

Step 15: Install OpenClaw

Install OpenClaw globally via npm:

sudo npm install -g openclaw

Step 16: Run the Setup Wizard

Run the onboarding wizard with the daemon flag to auto-start on boot:

openclaw onboard --install-daemon

Step 17: Accept the Terms

OpenClaw is powerful and has access to your system. Confirm you understand by selecting Yes:

Risk Confirmation

Step 18: Select Onboarding Mode

When prompted for Onboarding mode, select QuickStart:

Onboarding Mode

Step 19: Select Anthropic as Provider

Choose Anthropic as your model/auth provider:

Select Auth Provider

Step 20: Choose Auth Method

Select Anthropic token (paste setup-token) to use your Claude subscription:

Auth Method

Step 21: Paste Your Setup Token

Get your setup token by running claude setup-token on another machine with Claude CLI, or use an API key. Paste it here:

Paste Token

⚠️ WARNING: Make sure you paste the ENTIRE token! Copying and pasting from the terminal sometimes cuts off the end. Double-check that your token is complete before pressing Enter.

Give your token a name (or press Enter to use "default"):

Token Name

Step 22: Select Your Model

Choose your default model. We recommend anthropic/claude-opus-4-5 for the best capabilities:

Model List

You can also choose Sonnet for faster responses or Haiku for simple tasks.

Step 23: Configure Telegram

Select Telegram (Bot API) as your chat channel:

Select Telegram

To get a Telegram bot token:

1. Open Telegram and chat with @BotFather
2. Send /newbot and follow the prompts
3. Copy the token (looks like 1234567890:ABCdefGHIjklMNOpqrsTUVwxyz)

Step 24: Configure Skills (Optional)

OpenClaw will ask if you want to configure skills now. Select No to skip for now (you can always add skills later):

Skills Configuration

Step 25: Enable Hooks (Optional)

Hooks add extra functionality. Select Skip for now to keep things simple:

Enable Hooks

Step 26: Shell Completion (Optional)

You can optionally install shell completion for easier CLI usage. Select No to skip:

Shell Completion

Step 27: Hatch Your Bot!

Finally, choose how to launch your bot. Select Hatch in TUI (recommended) to start chatting immediately:

Hatch Bot

This opens the Terminal UI where you can chat with your AI assistant directly.

Step 28: Pair with Telegram

1. Open your Telegram bot and send any message
2. OpenClaw will reply with a pairing request
3. In your terminal, run:

openclaw pairing approve telegram <request-id>

🎉 Done!

Your AI assistant is now live on Telegram. Start chatting!

Cost Summary

Running OpenClaw on a t3.medium spot instance costs approximately:

Compare this to managed AI assistant solutions at $100+/month - you're saving 80%!

Next Steps

• Set up spot instance monitoring for termination alerts
• Configure additional models (Haiku for simple tasks)
• Enable web search and browser automation
• Connect Discord, Slack, or WhatsApp

Check the OpenClaw docs for more configuration options.

Questions? Join the OpenClaw Discord community!