Personal AI agent on Ubuntu

OpenClaw VPS hosting, your AI agent, always on

Give OpenClaw a private server of its own and let it work while you sleep. Message it from Telegram, WhatsApp or Slack, and it answers email, books meetings, browses the web and sorts your files. Your server is ready in minutes, and OpenClaw installs with a few commands.

From $7.99/mo billed yearly or $9.99 month to month

  • Full root access
  • DDoS protection
  • Dedicated IPv4
  • 7-day refund

the short answer

What is an OpenClaw VPS?

An OpenClaw VPS is a virtual private server that runs OpenClaw, the open-source personal AI agent, around the clock. You chat with it in apps like Telegram or WhatsApp, and it gets real work done: email, calendar, files and web research. On VPSPulse your server comes with Ubuntu and is online within minutes, with full root access, a dedicated IPv4 address and DDoS protection. You then install OpenClaw with its official installer in a few minutes. It suits anyone who wants a digital employee that keeps working when their own computer is off. Plans with 2 GB of RAM or more start at $7.99/mo billed yearly ($9.99 month to month), and you can choose from 17 plans in 12 cities across 12 countries.

From
$7.99/mo
Locations
12 cities
Plans
17
Server ready in
Minutes
Base system
Ubuntu 24.04

From checkout to your first task

  1. Pick a plan and a city

    Every Deploy button on this page opens the checkout with Ubuntu already selected. For a smooth setup we suggest 2 vCPU and 4 GB of RAM or more.

  2. Install OpenClaw in minutes

    Your server is online within minutes of payment. Connect over SSH and run the official installer: a few commands, and no hardware to buy or set up.

  3. Say hello from your chat app

    Connect Telegram, WhatsApp, Discord, Slack or Signal during onboarding, then send your agent its first job.

New to servers? Our step-by-step VPS setup guide walks you through your first login.

what openclaw does

From chatbot to digital employee

When you use ChatGPT or another AI chatbot, you ask questions and you get answers. It works a bit like a very clever search engine. OpenClaw changes that. You give it a goal, it takes action, shows you the result and keeps going until the job is done. It goes from chatbot to employee, and that is a big shift.

A typical chatbot

  • Answers the question you typed
  • Lives in a browser tab
  • Stops when you close the tab
  • You copy the answer and do the work

OpenClaw on your VPS

  • Works toward the goal you set
  • Lives in the chat apps you already use
  • Keeps running on its own server, 24/7
  • Reads email, browses, saves files and reports back

Open source, shaped your way

OpenClaw is an open-source project, so you can customize it any way you want: choose the AI model, add skills and decide which tools it may use. You don't need to open a website to talk to it, either. It lives where you already chat, so you can work with your agent from your phone.

Works where you already chat

  • Telegram
  • WhatsApp
  • Discord
  • Slack
  • Signal

Need fixed, step-by-step automations instead of an agent? Have a look at self-hosted n8n for workflow automation.

install guide

Install OpenClaw on your VPS

Your VPSPulse server arrives with a clean Ubuntu. OpenClaw is not pre-installed: you add it yourself with the official installer, which takes a few commands and a few minutes. That way you always get the current release straight from the OpenClaw project.

  1. Connect to your server

    Use the IP address and root password from your welcome email. Open a terminal on Windows, macOS or Linux and connect over SSH:

    ssh root@YOUR_SERVER_IP
  2. Run the official installer

    The installer checks your system, sets up Node.js if it is missing, installs OpenClaw and starts onboarding, where you pick your AI model and connect a chat app.

    curl -fsSL https://openclaw.ai/install.sh | bash

    Already have a current Node.js? Install with npm instead. The --install-daemon option sets OpenClaw up as a background service:

    npm install -g openclaw@latest
    openclaw onboard --install-daemon
  3. Check that it works

    These two commands look for configuration problems and confirm that OpenClaw's gateway is running:

    openclaw doctor
    openclaw gateway status

Good habits from day one

  • Keep the OpenClaw gateway on the server's local interface and open its dashboard through an SSH tunnel.
  • Give your agent only the accounts and permissions it really needs.
  • Back up OpenClaw's settings folder and update OpenClaw regularly.

Commands and options can change between releases, so take a quick look at the official OpenClaw installation docs before you start.

live plans

Choose your OpenClaw VPS plan

Every plan below has at least 2 GB of RAM and comes with Ubuntu, ready for OpenClaw. Prices are per month, and paying yearly saves 20%. For a smooth setup we suggest 2 vCPU and 4 GB of RAM or more. Start small and move to a bigger plan when you need it.

Shared CPU plans for OpenClaw

Shared vCPU cores at a friendly price. Plenty for a personal agent that talks to a hosted AI model, handles email and calendar, and runs scheduled jobs.

How much server does OpenClaw need?

These are our suggestions, not hard limits. OpenClaw itself is light when it uses a hosted AI model. A real browser and local models are what need the extra memory.

What you runWe suggest
Trying OpenClaw with a hosted AI modelA 2 GB RAM plan is enough for a first test.
A personal agent you use every day2 vCPU and 4 GB of RAM or more, for a smooth setup.
Browser automation, several chat apps4 vCPU and 8 GB of RAM.
Local AI models instead of a hosted oneMuch more memory: 16 to 32 GB of RAM or more. Our plans have no GPU, so only small models run well.
Lots of downloads, reports and backupsA Storage plan with a large disk.

Need a server for something else? See full VPS pricing or a plain Ubuntu VPS.

task ideas

What your OpenClaw agent can do for you

Here are four everyday jobs people hand to their agent, with the plan we would start each one on. You start each one with a chat message.

Inbox helper

Email · Telegram

Tell your agent what to do with your inbox. It reads new mail, picks out what matters and drafts replies in your tone for you to check.

  1. You ask in Telegram
  2. Reads new emails
  3. Sorts what matters
  4. Drafts the replies
  5. You approve, it sends

The server stays online, so your agent can check the inbox every morning, even when your laptop is closed.

Suggested plan: 4 vCPU · 6 GB RAM from $13.59/mo Deploy a server for Inbox helper

Morning briefing

Calendar · WhatsApp

Wake up to a short message with today's meetings, the news from sites you follow and anything that needs an answer.

  1. 7:00 every morning
  2. Checks your calendar
  3. Reads the sites you follow
  4. Writes a short summary
  5. Sends it to WhatsApp

Scheduled jobs run on time, because your agent lives on a machine that never sleeps.

Suggested plan: 4 vCPU · 6 GB RAM from $13.59/mo Deploy a server for Morning briefing

Web research and data extraction

Browser · Files

Give it a goal, such as comparing prices or collecting contact pages. The agent opens a browser on the server, visits the pages, pulls out the data and hands you a tidy file.

  1. You set the goal
  2. Opens a browser
  3. Pulls out the data
  4. Saves a CSV file
  5. Replies with a summary

A real browser needs memory, so we suggest 8 GB of RAM for this job. Downloads stay on your server's disk.

Suggested plan: 4 vCPU · 8 GB RAM from $17.59/mo Deploy a server for Web research and data extraction

Remote help for your projects

Discord · Slack · Terminal

Ask from your phone whether your website is up, what changed in a log file or how full the disk is. With root access, your agent can run the checks and report back.

  1. Ask in Slack or Discord
  2. Runs the checks
  3. Reads logs and status
  4. You approve any fix
  5. Reports back

It runs on its own server, so a wrong command can't touch your personal computer, and you can reset the server at any time.

Suggested plan: 4 vCPU · 6 GB RAM from $13.59/mo Deploy a server for Remote help for your projects

why a vps

Why run OpenClaw on a VPSPulse server

  • A safe sandbox for a powerful agent

    OpenClaw can read files, run commands and use your accounts. On its own server it stays apart from your personal computer and everything on it.

  • Online 24/7

    If you switch off your computer, a local OpenClaw stops. A server keeps your digital employee working day and night.

  • Easy to start fresh

    If something goes wrong, reinstall the server with a clean Ubuntu and set OpenClaw up again in minutes. No device to wipe by hand.

  • No hardware to buy

    No Mac mini or spare laptop needed. You pay a flat price from $7.99/mo, with no fee per message or task from us.

  • Full root access

    It's your Linux server. Add a browser, Python, Docker or anything else your agent needs.

  • DDoS protection and your own IPv4

    Every server includes DDoS protection and a dedicated IPv4 address.

  • 12 cities in 12 countries

    Put your agent close to you, or close to the services it talks to most.

  • Real help, low risk

    Support in 8 languages, card, PayPal, bank transfer or crypto, 20% off when you pay yearly, and a 7-day refund for new customers.

A VPS or a spare computer at home?

Both can run OpenClaw. A computer at home makes sense if you already own a strong one and want to run large local models. A VPS is the easier choice when you want your agent always on, kept apart from your personal machine, and nothing to buy.

ComparedOpenClaw on a VPSPulse VPSA spare computer at home
Upfront costNone, from $7.99/mo billed yearlyYou buy the device first
Online all dayYes, it runs in a datacenterOnly while it's on and your home internet works
Kept apart from your filesYes, it's a separate serverOnly if the device is used for nothing else
Starting freshReinstall the server, set up again in minutesWipe and reinstall by hand
Network addressDedicated IPv4, no router setupBehind your home router
Local AI modelsCPU only: small models on large RAM plansDepends on your hardware; a strong GPU runs bigger models

faq

OpenClaw VPS FAQ

What is OpenClaw?

OpenClaw is an open-source personal AI agent. With a chatbot such as ChatGPT, you ask questions and get answers. With OpenClaw you give it a goal instead: it takes action, shows you the result and keeps going until the job is done. That is why people call it a digital employee rather than a chatbot. You talk to it in the chat apps you already use, such as Telegram, WhatsApp, Discord, Slack or Signal. OpenClaw is an independent open-source project. VPSPulse doesn't make it; we give it a fast, always-on home.

How is OpenClaw different from ChatGPT?

ChatGPT answers; OpenClaw acts. It can read and answer email, browse websites, pull data out of pages, work with files and connect to your tools, then report back in your chat. Because it is open source, you decide which AI model it uses and what it is allowed to touch.

Why run OpenClaw on a VPS?

The first reason is security. OpenClaw is powerful, so running it on a separate server gives it a sandbox: if something goes wrong, you reset that server, not your own computer. The second is staying online. When your computer is off, a local OpenClaw stops, while a server keeps your agent online 24/7. The third is cost. Instead of buying a Mac mini or keeping an old laptop running, a VPS starts at $7.99/mo billed yearly.

Is OpenClaw pre-installed on VPSPulse servers?

No. Your server comes with a clean Ubuntu (you can choose 24.04) and is ready within minutes. You then install OpenClaw yourself with its official installer, which takes a few commands and a few minutes. This way you always get the current release straight from the OpenClaw project.

How do I install OpenClaw on a VPS?

Connect to your server over SSH as root and run the official installer: curl -fsSL https://openclaw.ai/install.sh | bash. It sets up Node.js if needed, installs OpenClaw and starts onboarding. If you already have a current Node.js, you can run npm install -g openclaw@latest followed by openclaw onboard --install-daemon instead. Then check everything with openclaw doctor and openclaw gateway status.

What server do I need for OpenClaw?

We suggest 2 vCPU and 4 GB of RAM or more for a smooth setup, for example 4 vCPU · 6 GB RAM (from $13.59/mo billed yearly). A 2 GB plan is fine for a first test with a hosted AI model. For browser automation or several chat apps, go for 8 GB of RAM. Start small and move to a bigger plan when you need it.

Can OpenClaw use local AI models instead of a hosted one?

Yes, but local models need far more memory than OpenClaw itself, and they run much faster on a GPU. Our plans run on CPU only, so local models are limited to small ones on plans with plenty of RAM. Most people connect OpenClaw to a hosted AI model with an API key, which asks very little of the server.

Which messaging apps can I connect to OpenClaw?

OpenClaw works with chat apps such as Telegram, WhatsApp, Discord, Slack and Signal. You connect them during onboarding or later, and then you can give your agent tasks straight from your phone.

Is OpenClaw free?

OpenClaw itself is open source and free to use. You pay VPSPulse for the server, and if you use a hosted AI model, that service bills you for what you use. VPSPulse never charges per message or per task.

Is it safe to give an AI agent access to my accounts?

Treat it like a new assistant and give it only the accounts and permissions it needs. Running it on its own VPS keeps it away from your personal files. Keep OpenClaw's gateway on the server's local interface, open the dashboard through an SSH tunnel, update OpenClaw regularly and back up its settings.

How fast is my server ready, and how can I pay?

Your server is online within minutes of payment. You can pay by card, PayPal, bank transfer or cryptocurrency, monthly or yearly (yearly saves 20%), and new customers can ask for a refund within 7 days.

Where are your OpenClaw servers located?

You can run OpenClaw in 12 cities across 12 countries: Helsinki (Finland), Paris (France), Frankfurt (Germany), Reykjavik (Iceland), Milan (Italy), Amsterdam (Netherlands), Warsaw (Poland), Madrid (Spain), Zurich (Switzerland), Istanbul (Turkey), London (United Kingdom), New York (United States).

Ready to give your AI agent a home?

Get an Ubuntu server from $7.99/mo billed yearly, install OpenClaw in a few minutes and keep your digital employee online around the clock. New customers have 7 days to change their mind.

Prefer fixed, step-by-step automations? Try self-hosted n8n or browse all one-click apps.