How to Install OpenClaw and Deploy on Fluence Cloud

Fluence
[10 min read]
Install OpenClaw on Fluence Cloud

Running OpenClaw on a virtual server is a practical way to move from a local test setup to something that stays online and accessible. Instead of depending on your laptop, terminal session, or home network, you run the OpenClaw gateway on a dedicated Linux VM that remains available in the background.

In this guide, you will learn how to deploy and run you own OpenClaw AI agent on Fluence Cloud from start to finish. The walkthrough covers provisioning the VM, connecting over SSH, installing OpenClaw, completing onboarding, verifying that the gateway is running correctly, and then choosing how you want to use it afterward. 

If you want a browser-based experience, the guide also shows how to connect to the OpenClaw dashboard securely from your laptop using SSH tunneling.

What this guide covers

This guide shows how to:

  1. Provision a Linux VM on Fluence.
  2. Connect to it from your laptop.
  3. Install OpenClaw.
  4. Run OpenClaw’s onboarding so the gateway is installed as a background service.
  5. Verify that OpenClaw is successfully installed.
  6. Choose how to use it next:
    • Option A: use a messaging app such as Telegram as the main chat interface.
    • Option B: use the OpenClaw dashboard securely from your laptop.

For security, this guide keeps the OpenClaw gateway on loopback and uses SSH tunneling for dashboard access instead of exposing the dashboard directly to the internet. OpenClaw’s VPS guidance recommends that as the secure default, and its onboarding docs recommend keeping token auth enabled even on loopback.

Before you start

You need four things:

  • A Fluence account with enough balance to launch a VM. Fluence reserves 2 days of rent when you deploy an instance.
  • An SSH key pair. Fluence requires at least one public SSH key when you launch a VM.
  • An API key for the AI model provider you plan to use, such as OpenAI or Anthropic. OpenClaw onboarding will ask for it.
  • A Mac or Linux laptop with terminal access, so you can connect to the VM over SSH.

Step 1: Add funds to your Fluence balance

In Fluence Console, go to the Billing page and top up your balance before you launch the VM. Fluence’s docs say VM rent is paid from your Fluence balance, and deployment reserves two days of rent up front.

Step 2: Create an SSH key on your laptop

If you already have an SSH key you want to use, you can skip this step.

If not, run this on your laptop’s Terminal:

ssh-keygen -t ed25519 -C "fluence-openclaw"

cat ~/.ssh/id_ed25519.pub

Copy the full public key output. Go to Settings, click SSH keys, click Add SSH Key and paste your SSH public key in the input box.

image

Step 3: Create the VM in Fluence Console

In Fluence Console, open Public Cloud, go to Compute, and create a new instance.

Choose a basic instance to install OpenClaw and deploy within a few minutes

Follow this sequence in the Console:

  1. Click “Create new VM”
  2. Choose a location.
  3. Choose a configuration. Fluence CPU Cloud starts from 2 vCPUs and 4 GB RAM.
  4. Choose OS image. The pre-defined OS is Ubuntu 22.04, so leave it as it is.
  5. Rent a public IPv4 address (to access the instance via SSH).
  6. Set an instance name.
  7. Add or select your public SSH key.
  8. Review the summary and click “Create VM” to launch the instance.

Review your VM status under the “Compute” section of the console. Once the VM status becomes Active, and you can see its public IP address (you will need to copy-paste it in the next step).

Step 4: Connect to the VM from your laptop

From your laptop terminal, connect using SSH:

ssh ubuntu@YOUR_VM_PUBLIC_IP

Accept the host fingerprint the first time.

If the Ubuntu image uses a different default username in your environment, use that username instead.

Once the SSH connection is successful, your terminal prompt changes and you are now inside the VM.

Step 5: Update the VM

Run these commands inside the VM:

sudo apt update && sudo apt upgrade -y

sudo apt install -y curl ca-certificates

This prepares the VM for the OpenClaw installer.

If the commands finish without errors and you return to the shell prompt, you can proceed to the next step to install OpenClaw on your VM.

Step 6: Install OpenClaw on the VM

For a clean, predictable walkthrough, install OpenClaw without launching onboarding automatically:

curl -fsSL https://openclaw.ai/install.sh | bash -s -- --no-onboard

OpenClaw’s install docs say this is the recommended installer path for macOS/Linux/WSL2, and that the installer handles Node automatically if needed. They also document the –no-onboard option for installing first and running onboarding separately afterward.

Now verify the CLI is available:

openclaw --version

You will see an OpenClaw version number.

If you get “command not found,” OpenClaw’s install docs say to check whether $(npm prefix -g)/bin is in your PATH, and add it if needed:

node -v

npm prefix -g

echo "$PATH"

export PATH="$(npm prefix -g)/bin:$PATH"

echo 'export PATH="$(npm prefix -g)/bin:$PATH"' >> ~/.bashrc

source ~/.bashrc

openclaw --version

Step 7: Run OpenClaw onboarding

Now start OpenClaw’s setup wizard:

openclaw onboard --install-daemon

OpenClaw’s onboarding docs say this flow configures your model provider, workspace, gateway settings, optional chat channels, installs a daemon, and runs a health check. On Linux, it installs a systemd user unit and attempts to enable lingering so the gateway can keep running after logout.

When the wizard asks questions, use these choices:

Model provider and API key

Choose your preferred provider and paste your API key.

This is required because OpenClaw needs a model backend to work.

Workspace

Accept the default workspace unless you have a reason to change it.

OpenClaw’s CLI setup reference says the default workspace is ~/.openclaw/workspace.

Gateway settings

Use the secure defaults:

  • keep the gateway bound to loopback/local only
  • keep token auth enabled
  • accept the default port unless you have a reason to change it

OpenClaw’s CLI reference says onboarding prompts for port, bind, and auth mode, and recommends keeping token auth enabled even for loopback. Its getting-started guide says the gateway should come up on port 18789 in the standard quick setup.

Channels

When the wizard asks about channels or chat apps, choose Skip or leave channels unconfigured for now.

That keeps this guide focused on installation only.

Daemon install

Say Yes to install the daemon.

On Linux, this gives you managed startup via a systemd user service.

Health check

Let the wizard complete its health check.

OpenClaw says onboarding starts the gateway if needed and runs openclaw health.

What success looks like: onboarding finishes without a blocking error and returns you to the shell prompt.

Step 8: Confirm OpenClaw is installed successfully

Run these checks on the VM:

openclaw doctor

openclaw gateway status

openclaw status --deep

systemctl --user status openclaw-gateway.service

OpenClaw’s install and troubleshooting docs list these as the standard ways to verify the CLI, validate configuration, confirm the gateway is running, and perform deeper health checks.

You have reached the end of the installation process when all of the following are true:

  • openclaw –version works
  • openclaw doctor does not show a blocking setup problem
  • openclaw gateway status shows the gateway is running
  • the systemd user service is active

That is the clean stopping point for “OpenClaw is successfully installed on the VM.”

If the service does not stay up after logout, OpenClaw’s onboarding docs say Linux installs use a systemd user unit and the wizard attempts loginctl enable-linger <user> so the service can survive logout. If needed, you can run:

sudo loginctl enable-linger $USER

systemctl --user enable --now openclaw-gateway.service

After installation: choose your main chat interface

At this point, OpenClaw is installed and running on your Fluence VM.

You now have two ways to use it:

Option A: Use a messaging app such as Telegram

For this path, read Fluence’s step-by-step tutorial on connecting OpenClaw to Telegram. That article is the right place for the bot-token and pairing flow.

Option B: Use the OpenClaw dashboard from your laptop

This is the path covered below.

Step 9: Connect to the OpenClaw dashboard securely from your laptop

Because OpenClaw recommends keeping the gateway on loopback for VPS deployments, the safe pattern is:

  • do not expose the dashboard port publicly
  • keep only SSH open on the VM
  • use an SSH tunnel from your laptop to the VM

That recommendation comes directly from OpenClaw’s VPS guidance. Its getting-started guide also says the dashboard is served by the gateway and that the standard setup listens on port 18789.

From your laptop, run:

ssh -N -L 18789:127.0.0.1:18789 ubuntu@YOUR_VM_PUBLIC_IP

What this does:

  • -L 18789:127.0.0.1:18789 forwards your laptop’s local port 18789 to the VM’s local OpenClaw port 18789
  • -N tells SSH to keep the tunnel open without starting a remote shell

Leave that terminal window open.

Now, on your laptop, open this in a browser:

http://127.0.0.1:18789

If the dashboard loads, your secure tunnel is working.

As an alternative check from the VM itself, OpenClaw’s getting-started guide also documents:

openclaw dashboard

On a local desktop Linux session, that opens the Control UI in a browser. In a headless server setup, the onboarding docs say the wizard prints SSH port-forward instructions instead of opening a browser when no GUI is detected.

If the prior steps were done correctly, the OpenClaw dashboard will open in your laptop browser and prompt you for access using the local, tunneled connection.

Troubleshooting

If something is not working, start with these commands on the VM:

openclaw status

openclaw status --deep

openclaw gateway status

openclaw logs --follow

openclaw doctor

OpenClaw’s troubleshooting docs recommend exactly these commands as the first checks for install, runtime, and gateway issues.

The most common issues in this workflow are:

  • the VM is not yet in Active state on Fluence
  • the wrong SSH key was attached during VM creation
  • openclaw is installed but not in PATH
  • onboarding finished incompletely
  • the gateway is not running as a service
  • the SSH tunnel command is not currently open on your laptop

Summary

By the end of this guide, you have:

  • a Fluence VM running Ubuntu
  • OpenClaw installed on that VM
  • the OpenClaw gateway configured and running as a background service
  • a secure way to use the OpenClaw dashboard from your laptop without exposing it publicly

That gives you a clean base article framework: install first, then branch into either messaging app usage or dashboard usage depending on the reader’s preferred interface.

Start with a basic instance at $9 a month to install and deploy your own OpenClaw AI agent

Install openclaw now

Conclusion

At this point, you have a working OpenClaw installation running on a Fluence Virtual Server. You created the VM, connected to it from your laptop, installed OpenClaw, completed onboarding, and confirmed that the gateway is running as a background service.

From here, you can choose the interface that fits your workflow best. You can continue with a messaging-based setup such as Telegram, or you can use the OpenClaw dashboard securely from your laptop through an SSH tunnel. Either way, you now have a clean and reliable foundation for running OpenClaw on an always-on server instead of a local machine.

This setup gives you a strong starting point for testing, personal use, and early production-style deployments while keeping the infrastructure lightweight and manageable.

Similar articles

To top