Guaranteed Results
Hi, I'm Phillip Neiman. I help individuals and businesses deploy self-hosted AI assistants on AWS quickly and reliably. I guarantee you'll have a working AI assistant you can text — or you don't pay. I've navigated the gotchas to get your OpenClaw bot online. Learn more →
Stop paying $20-200/month for ChatGPT subscriptions. Deploy your own AI assistant that's smarter, private, and costs pennies per conversation. I guarantee you'll have it working.
Your data never leaves your control. No corporate servers analyzing your conversations.
Access via Telegram, WhatsApp, or Discord. Your AI assistant follows you everywhere.
Claude Sonnet 4.5 costs ~$0.05 per conversation. No subscriptions, no limits.
Automate tasks, remember context forever, integrate with your tools. Not just a chatbot.
I'll show you exactly how. No DevOps experience required. Guaranteed working or your money back.
Get Started →Your AI assistant runs on battle-tested infrastructure
You message @BotFather to create your bot
Ubuntu server running OpenClaw 24/7
Sonnet 4.5 or Opus 4.6 powers responses
Intelligent replies back to you
Cloud hosting
Linux OS
Agent framework
@BotFather
Sonnet 4.5 / Opus 4.6
Runtime
This is what your AI assistant looks like once deployed
Follow step-by-step instructions written during real installations. No guesswork, just clear directions that work. Guaranteed.
Every issue documented here was actually encountered and solved. You get solutions to problems you will actually face.
You'll have a working bot you can text, or you don't pay. Save time and money with guidance from someone who's done this before.
Real problems. Real solutions. No fluff.
The #1 issue everyone hits: JavaScript heap out of memory during onboarding. Complete fix with instance resizing or swap space.
SSH timeouts after resizing, host key verification failures, and permission errors on key files. All documented with fixes.
Finding port settings, understanding CIDR notation, and dealing with AWS warnings. Clear explanations of what's happening.
Gateway not started, wrong bot username, token mismatches. Every reason your bot might be silent and how to diagnose it.
Free tier workarounds, Elastic IP setup, and spending alerts configuration. Keep costs under control from day one.
Running as a systemd service, auto-restart configuration, and log monitoring. Make your assistant reliable for daily use.
A complete, battle-tested guide to deploying OpenClaw on AWS with Telegram integration. Written during real installations with every gotcha documented. Guaranteed to work or your money back.
OpenClaw is an open-source, self-hosted AI agent that connects to messaging platforms like Telegram, WhatsApp, Discord, Slack, and more. It acts as a gateway between you and an AI model (like Claude or GPT), letting you interact with a powerful AI assistant directly from your phone.
This guide walks you through deploying OpenClaw on an AWS EC2 instance with Telegram as your primary interface. It was written during an actual installation, so it includes real troubleshooting scenarios and gotchas you are likely to encounter.
30–45 min: First-time setup (including AWS account)
15–20 min: If you already have AWS + API keys ready
An EC2 instance is a virtual computer you rent from Amazon. It runs 24/7 so your bot is always available.
| Setting | Value |
|---|---|
| Name | openclaw-bot |
| AMI | Ubuntu 24.04 LTS |
| Instance type | t3.small (2GB RAM) |
| Key pair | Create new → download .pem file |
| Storage | 8 GB gp3 |
Security groups control what traffic can reach your instance. You need two ports open:
| Port | Protocol | Source | Purpose |
|---|---|---|---|
| 22 | TCP | Your IP /32 | SSH access |
| 18789 | TCP | Your IP /32 | OpenClaw Dashboard |
98.45.123.67/32ssh -i C:\Users\YourName\Downloads\your-key.pem ubuntu@YOUR_PUBLIC_IPssh -i ~/Downloads/your-key.pem ubuntu@YOUR_PUBLIC_IPyes (the full word) and press Enter. Just pressing Enter will fail.
Mac/Linux:
chmod 400 ~/Downloads/your-key.pemWindows:
icacls "C:\Users\YourName\Downloads\your-key.pem" /inheritance:r /grant:r "%USERNAME%:R"curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash -
sudo apt install -y nodejsmkdir -p ~/.npm-global
npm config set prefix '~/.npm-global'
echo 'export PATH=~/.npm-global/bin:$PATH' >> ~/.bashrc
source ~/.bashrcnpm install -g openclaw/newbotopenclaw onboard| Prompt | What to Do |
|---|---|
| Bot Token | Paste token from BotFather |
| Package manager | Select npm |
| Skills | Skip for now |
| API keys | Enter Anthropic API key |
| Hatch | Select "Hatch in TUI" |
openclaw pairing approve telegram YOUR_CODEsudo tee /etc/systemd/system/openclaw-gateway.service << 'EOF'
[Unit]
Description=OpenClaw Gateway
After=network.target
[Service]
Type=simple
User=ubuntu
WorkingDirectory=/home/ubuntu
ExecStart=/home/ubuntu/.npm-global/bin/openclaw gateway start --foreground
Restart=always
RestartSec=10
Environment=PATH=/home/ubuntu/.npm-global/bin:/usr/local/bin:/usr/bin:/bin
[Install]
WantedBy=multi-user.target
EOFsudo systemctl daemon-reload
sudo systemctl enable openclaw-gateway
sudo systemctl start openclaw-gateway| Command | What It Does |
|---|---|
sudo systemctl status openclaw-gateway |
Check status |
sudo systemctl restart openclaw-gateway |
Restart |
sudo journalctl -u openclaw-gateway -f |
View live logs |
FATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memoryYour instance doesn't have enough RAM. t3.micro only has 1GB.
sudo fallocate -l 2G /swapfile
sudo chmod 600 /swapfile
sudo mkswap /swapfile
sudo swapon /swapfile
echo '/swapfile none swap sw 0 0' | sudo tee -a /etc/fstabStopping/starting changes your public IP.
You pressed Enter instead of typing "yes".
Run SSH command again. Type yes (full word) when prompted.
openclaw gateway status| Component | Cost |
|---|---|
| AWS EC2 t3.small | ~$15/month |
| OpenClaw | Free |
| Telegram | Free |
| AI API (Sonnet 4.5, casual) | ~$10–30/month |
| Total | ~$25–45/month |
You will have a functional AI assistant responding to your messages. Not a half-working tutorial. Not "almost there." A working bot you can text right now.
If you follow my guide and can't get it working, or hire me and I don't deliver — you don't pay. Simple as that.