The $12/Month VPS Setup That Ran My Business While I Slept

Everyone in the OpenClaw community talks about hardware. Raspberry Pi 5 builds. Mac Mini M4 workstations. Rack-mounted home servers with 128GB RAM.

That’s fine if you’re a tinkerer. But if you just want your business to run while you sleep, there’s a cheaper, faster path: a $12/month cloud VPS.

Here’s what one is actually doing for a real solopreneur right now: handling WhatsApp inquiries, qualifying leads, booking appointments, and sending a morning summary — all overnight, zero human involvement.

Let’s break down how it works and how to replicate it.

Why VPS Over Raspberry Pi or Mac Mini?

Raspberry Pi is great for home labs. Mac Mini is great if you already have one and want a local AI workstation. But both have real tradeoffs:

Raspberry Pi is cheap but runs on your home internet. If your connection flickers, your agent goes dark.
Mac Mini requires you to own a $599+ machine, keep it powered, and manage home networking.
VPS gives you a Linux server in a proper datacenter — redundant power, proper uptime, managed networking — for $10–15/month.

For a business that needs 24/7 availability to respond to leads and handle customer interactions, a VPS wins on reliability.

The Setup: What a $12/Month Server Can Actually Do

Here’s the overnight workflow that has been circulating on X:

A potential customer sends a WhatsApp message at 11pm asking about pricing
OpenClaw’s WhatsApp integration picks it up within seconds
The agent reads the message, checks a Google Sheet with pricing tiers and FAQs
It sends a personalized, on-brand response with a Calendly booking link
The customer books a call for tomorrow morning
At 7am, the owner wakes up to a summary: “4 inquiries, 2 leads qualified, 1 appointment booked, 1 follow-up needed”

No missed leads. No late-night phone checking. No context-switching.

This is not a demo — it’s running live. And the infrastructure cost is $12/month.

Picking the Right VPS for OpenClaw

Not all cheap VPS providers are equal. OpenClaw with a lightweight local model needs RAM above all else. Here’s what to look for:

Minimum viable specs:

4GB RAM (2GB will work but you’ll hit limits fast)
2 vCPUs
20GB SSD
Ubuntu 22.04 or 24.04 LTS

Providers worth checking:

Hetzner Cloud (CAX11 ARM64) — €3.79/month in Europe, fast storage, excellent uptime. Best bang-for-buck.
DigitalOcean Basic Droplet — $12/month for 2GB RAM, 1 vCPU. Slightly pricier but US-based and beginner-friendly.
Vultr High Frequency — competitive pricing, multiple datacenter locations.
OVHcloud Starter — €3.50/month, solid for budget deployments.

If you’re running OpenClaw with cloud models (Claude, GPT-4o) rather than local models, you can drop to 2GB RAM easily — you’re just running the agent orchestration layer, not the model itself.

Step-by-Step: Deploying OpenClaw on a VPS

1. Spin Up Your Server

Choose Ubuntu 24.04 LTS. Enable SSH key authentication (skip password auth). Pick a datacenter region close to you or your customers.

2. Install OpenClaw

SSH into your server and run the official installer from openclaw.dev — it handles Node.js, the CLI, and systemd service setup in about 3 minutes.

3. Configure Your Gateway

openclaw gateway start
openclaw gateway status

The gateway is what keeps your agent alive 24/7 — the persistent process that listens for triggers (messages, cron jobs, webhooks) and wakes your agent when needed.

4. Set Up Your Secrets

Your VPS needs API keys for whatever integrations you’re using. Store them in a locked-down ~/.secrets file with chmod 600. Include your LLM API key plus any WhatsApp, Calendly, or Google credentials.

5. Create Your Agent

This is where the actual business logic lives. A minimal lead-qualification agent needs:

SOUL.md — the persona and response style
AGENTS.md — what the agent does and doesn’t do
A memory folder — so context persists across conversations
Cron entries — for scheduled tasks like morning summaries

6. Connect Your Channels

OpenClaw’s channel plugins handle the integration layer. For WhatsApp, Slack, Discord, email — there are plugins for each.

The key insight: OpenClaw doesn’t become a WhatsApp bot. It’s an orchestration layer. It receives a message, thinks about it, calls the right tools, and responds. You could swap WhatsApp for SMS or email and the agent logic stays the same.

The Morning Summary Workflow

This is the part most people sleep on (literally). A cron-triggered summary changes how you start your day.

Configure a daily 7am task that tells your agent to summarize overnight activity: leads received, responses sent, appointments booked, anything needing human review.

Every morning your agent pushes you a structured brief before you’ve had coffee. You know exactly what happened while you were offline and what needs your attention. It’s the closest thing to having an overnight assistant who actually shows up.

What to Automate First

If you’re starting fresh, here’s the order that delivers ROI fastest:

Lead response — First responder wins. Automate your initial reply to any inquiry channel.
FAQ handling — Give your agent a knowledge base. 80% of questions are the same 20 questions.
Appointment booking — Integrate Calendly or Cal.com. Your agent qualifies and books; you just show up.
Morning brief — Daily summary of what happened overnight.
Follow-up sequences — If a lead doesn’t book, trigger a follow-up at 24h and 72h.

Build incrementally. You don’t need all five on day one.

Real Costs, Honest Math

Here’s what this actually runs:

VPS (Hetzner CAX11): ~$5/month
OpenClaw (self-hosted, free): $0
Claude API (light usage): ~$3–8/month
WhatsApp Business API: ~$0 (first 1,000 messages free)
Total: $8–13/month

If you’re closing one extra lead per month because you responded at 11pm instead of 9am the next day, this pays for itself in the first week.

The Limits (Be Honest With Yourself)

This setup is not magic. A few things to be clear about:

Complex negotiations need humans. Your agent handles qualification and booking, not closing.
Context windows matter. Long conversation threads need memory management — build that in from day one.
API costs scale. If you go viral and get 500 inquiries overnight, your Claude bill will reflect it. Set usage limits.
Prompt quality is everything. A bad SOUL.md turns your agent into a liability. Write it carefully.

Start Small, Scale When It Works

The biggest mistake in automation is trying to build everything at once. Start with one workflow — lead response or morning summary. Run it for two weeks. See what breaks. Fix it.

Once you trust it, add the next layer.

A $12/month VPS is cheap enough that you can experiment without commitment. If it doesn’t work, shut it down. If it does, you just bought back hours of your life every week for the cost of two coffees.

That’s the real value proposition — not the technology, but the time.

Want to go deeper? OpenClaw’s docs have the full integration guides. The VPS setup is easier than most people think.