What happens behind the scenes
When you start a free trial or complete payment, OneClickClaw kicks off a fully automated provisioning pipeline that transforms a bare virtual private server into a production-ready AI agent host. This section explains every step of that process in plain language so you know exactly what is happening while you wait.
You do not need to understand any of this to use OneClickClaw. It all happens automatically. But if you are curious about what is going on while you wait, read on.
Note
If provisioning takes longer than 40 minutes, please open a support ticket. Our team is automatically alerted when provisioning exceeds the expected time window, but a support ticket ensures we prioritize your deployment.
Provisioning timeline
An OpenClaw build takes about 25 minutes and a Hermes Agent build five to seven minutes. Both go through three stages:
Server creation
A brand-new virtual server is created in our EU datacenter in Denmark. The server is provisioned with the exact CPU, RAM, and SSD specifications matching your plan (for example, 2 vCPU and 4 GB RAM for the Starter tier). By the end of this step, a fresh Ubuntu installation is running and accessible over the network.
Software installation (the longest stage)
Our bootstrap script runs on the new server and installs everything your agent needs:
- Your agent: OpenClaw with Node.js, the runtime that powers it, plus its chat add-ons and a check of its own settings; or Hermes Agent with its own Python setup
- Nginx (the web server that handles encrypted traffic)
- SSL certificates (so all connections are secure)
- Process manager (keeps your agent running 24/7, restarts it if it crashes)
The script also hardens security by configuring the firewall, disabling password login, and setting up SSH key authentication.
Starting your agent and final checks
Your agent is started as a managed service, its web address gets its SSL certificate, health monitoring is enabled, and the server sends a secure callback signal to OneClickClaw confirming it is ready. At this point your server shows "Running" in your dashboard, we email you, and the build page shows the button that opens your agent. OpenClaw starts on its Model Setup page.
Creating your server
✓ Complete - VPS created in Denmark (DK-DC1)
Preparing the system
✓ Complete
Downloading OpenClaw / Installing Hermes Agent
✓ Complete
Starting your agent
⟳ In progress
Securing your web address, Final checks
○ Waiting
What each status means
During provisioning, your server passes through several status states. Here is what each one means:
| Status | What It Means | Action Required |
|---|---|---|
| Provisioning | Your server is being created and configured. The automated pipeline is running. | None. Wait for completion (about 25 minutes for OpenClaw, five to seven for Hermes Agent). |
| Running | Your server is fully configured and your agent is active, waiting for its AI model. | OpenClaw: open your agent and connect a model on its Model Setup page. Hermes Agent: connect your AI from the AI Model tab, unless you already chose it on the build page. |
| Upgrading | You chose to move your agent to a newer version, and the upgrade is running. It takes about half an hour. | None. Your agent is back when the upgrade ends. |
Note
On the build page you also see the named steps of the build, such as "Preparing the system" or "Securing your web address". These are normal parts of the main stages.
What can go wrong
Provisioning failures are uncommon, but when they do occur they typically fall into a few categories:
- Infrastructure delays. The datacenter may be experiencing unusually high demand, which can slow down server creation. These delays are temporary and usually resolve within minutes. The provisioning pipeline automatically waits and retries.
- Network issues. If a software repository is temporarily unreachable, package installation may fail. Our bootstrap script includes automatic retries with exponential backoff for transient network errors.
- A server that does not finish. In rare cases a new server stops partway. A server that is still working is given extra time, up to 40 minutes. If it stopped, our system builds a new server for you by itself, and the build page shows "Rebuilding".
In all cases, OneClickClaw monitors the provisioning pipeline and takes corrective action automatically. If automatic recovery fails, our support team is alerted and intervenes manually. You are never charged for a deployment that does not complete successfully.
Warning
An OpenClaw build normally takes about 25 minutes and a Hermes Agent build five to seven. If your server still shows "Provisioning" after 40 minutes, contact support and we will resolve it. You will not be charged for the failed deployment.
How we know when your server is ready
Once the bootstrap script finishes installing and configuring everything on your new server, the server sends a secure callback signal to OneClickClaw. Think of it like a notification from the server saying, "I am set up and ready to go." This callback includes a cryptographic token that proves the signal is authentic and not coming from an impersonator.
When OneClickClaw receives this callback, it updates your dashboard status from "Provisioning" to "Running", emails you, and shows the button that opens your agent on the build page. OpenClaw starts on its Model Setup page. If the callback does not arrive within the expected time window, our system automatically checks on the server and, if needed, retries the final configuration step.
This two-layer approach (callback plus automatic health check) ensures that no deployment gets stuck in a "Provisioning" state indefinitely. Even if the callback fails due to a network blip, the health check system catches it within minutes and completes the process.
Tip
You do not need to understand or interact with the callback system. It operates entirely behind the scenes. All you need to do is wait for the "Running" status or our email, then open your agent and connect your AI.