There's a new side hustle making the rounds in tech circles: on-site OpenClaw installation services, charging 500 yuan (roughly $70 USD) per session. Freelancers show up at a client's office, deploy an AI agent on their infrastructure, configure the model and messaging channels, and walk away in under two hours. The demand is real — and it tells us something important about where AI deployment is headed.
If OpenClaw is open-source and free, why would anyone pay $70 for someone to install it? Three reasons:
Most small business owners and e-commerce sellers aren't developers. They've heard about AI agents, they know they need one, but the gap between "I want this" and "I have this running" feels enormous. SSH, API keys, daemon mode, webhook configuration — it's a foreign language.
Even for technically capable people, a first-time deployment involves reading documentation, troubleshooting edge cases, and configuring channels. That's 2–4 hours of focused work. For a busy store owner, paying $70 to skip that and get a working agent in 90 minutes is a no-brainer.
There's a psychological comfort in watching someone set it up, test it, and confirm it works before they leave. The client knows it's running correctly because they saw it happen.
Here's the thing — the entire reason on-site installation is viable as a business is that people don't realize how easy cloud deployment has become. With the right template, deploying OpenClaw is genuinely a 30-minute, zero-code process.
Tencent Cloud Lighthouse offers a pre-configured OpenClaw application template that eliminates virtually all the friction:
No SSH required for initial setup. No manual dependency installation. No Docker configuration. The template handles everything.
Let's demystify the process. Here's the complete workflow that on-site installers follow:
They create a Lighthouse instance with the OpenClaw template. This takes about 60 seconds.
They SSH in (or use OrcaTerm from the browser) and run:
# Launch the configuration wizard
clawdbot onboard
# Enter the client's LLM API key
# IMPORTANT: Never hard-code API keys in scripts or share them in chat logs
export OPENCLAW_API_KEY="sk-client-key-here"
The wizard walks through model selection, persona configuration, and basic settings. Time: 5 minutes.
They set up the client's preferred channel — usually Telegram or WhatsApp:
# Channel configuration happens during onboarding or via:
# clawdbot onboard
# Follow the prompts for channel selection and token input
Detailed guides for each platform:
They install the skills the client needs:
# From the OpenClaw chat interface:
# "Please install a skill for me using Clawhub; its name is ecommerce-cs-assistant."
# "Please install a skill for me using Clawhub; its name is agent-browser."
For the full skill guide: Installing OpenClaw Skills.
This is the step that ensures the agent runs 24/7:
loginctl enable-linger $(whoami)
export XDG_RUNTIME_DIR=/run/user/$(id -u)
clawdbot daemon install
clawdbot daemon start
clawdbot daemon status
They send test messages through the connected channel, verify the agent responds correctly, and hand the client a one-page cheat sheet for basic maintenance.
Total time: 60–90 minutes. Total difficulty: moderate for a first-timer, trivial for anyone who's done it once.
The on-site installation trend reveals a broader market truth: the demand for AI agents far outstrips the supply of people who know how to deploy them. This gap creates opportunities:
If you're reading this article, you're already technical enough to skip the installer. The complete deployment tutorial is here: OpenClaw Deployment Tutorial.
The process is:
clawdbot onboard.That's it. No on-site visit required. No $70 fee.
The on-site installation trend is a leading indicator. It means mainstream adoption of AI agents is happening right now — not in some hypothetical future, but in small shops, solo businesses, and e-commerce stores today. The tools have caught up to the demand.
Whether you hire an installer or do it yourself, the important thing is to get your agent running:
The $70 is better spent on your first month of API tokens. Trust me.