If you’re serious about running an always-on AI worker, at some point you’ll want to install Hermes Agent on Hostinger instead of your own laptop.
I’ve had Hermes Agent running locally on my Mac for months, and it’s genuinely one of the best free, open-source agent frameworks out there.
But local has a ceiling.
Long-running goal loops, background computer use, and Kanban swarms all fight your machine for RAM and CPU the second you close the lid or open a video call.
Mike Power, one of the builders in the AI Profit Boardroom community, has been telling people the same thing for weeks: move Hermes Agent to a VPS like Hostinger and the performance issues disappear.
So in this guide I’m walking through exactly how I install Hermes Agent on Hostinger, why it’s worth doing, and what to watch out for.
What Is Hermes Agent?
Hermes Agent is a self-improving, open-source AI agent framework built by Simon Shemmans.
It’s free to run, it’s flexible, and it’s become the centrepiece of a lot of the AI agent content I make on this site.
Under the hood it can run on models like MiniMax-M2.7 or Gemma 4 through Ollama, and it currently ships as release line v0.13, nicknamed Tenacity.
It has two main surfaces you’ll actually touch day to day: Hermes Desktop, which is a fully offline client, and Hermes Workspace, which handles Kanban-style swarms of agents plus computer use.
Officially, there are four supported ways to install it: Unix, PowerShell, Ollama, and Hostinger.
That last one isn’t really a separate install method so much as it is “install the Unix version, but on a VPS instead of your own machine” — and that distinction matters, because it changes everything about reliability.
Old Way vs New Way: Running Hermes Agent Locally vs on Hostinger
Here’s the honest comparison, because I ran it both ways before switching.
| Old Way: Local Mac / PC | New Way: Hostinger VPS |
|---|---|
| Agent competes with every other app for RAM and CPU | Agent gets a dedicated environment with resources reserved just for it |
| Closing your laptop lid kills long-running goal loops | Runs 24/7, whether your laptop is open, closed, or on a plane |
| Fan noise and heat during long agent runs | Runs quietly on someone else’s rack, not your desk |
| Every teammate sets up their own local environment, differently | One documented environment everyone connects to the same way |
| Scaling up means buying a new machine | Scaling up means upgrading a VPS plan in a dashboard |
None of that means local is wrong — for quick tests and small tasks, running Hermes Agent on your own machine is still the fastest way to get started.
But the moment you want it running while you sleep, or want a client or teammate to hit the same instance, you want it on a VPS.
🔥 Want the exact setup I used?
Inside the AI Profit Boardroom I’ve got a full step-by-step section on deploying Hermes Agent, plus weekly coaching calls and 3,600+ members comparing notes on their own installs.
What You Need Before You Install Hermes Agent on Hostinger
You don’t need to be a sysadmin to do this, but you do need a few things ready first.
- A Hostinger VPS plan running Ubuntu — a KVM plan with at least 4GB RAM is a sensible starting point.
- SSH access to that VPS, which Hostinger gives you from the hPanel dashboard the moment the server provisions.
- An API key for whichever model you plan to call, or enough RAM headroom if you’re running MiniMax-M2.7 or Gemma 4 locally through Ollama on the server itself.
- Ten to fifteen minutes, because this is genuinely a short setup once the server is live.
Step-by-Step: Installing Hermes Agent on a Hostinger VPS
This is the same order I follow every time I spin up a fresh box.
Step 1: Provision the VPS.
Log into Hostinger, choose a VPS plan, and pick an Ubuntu LTS template so you’re on a clean, well-supported Linux base.
Step 2: SSH into the server.
Hostinger gives you the IP address and root credentials in hPanel — connect with a standard SSH command from your terminal.
Step 3: Update the box and install the basics.
Run your package manager’s update command, then install git, build tools, and Node.js, since Hermes Agent needs these to run its install scripts.
Step 4: Run the Unix install method.
Since Hostinger’s servers are Linux, you use the same Unix install path Hermes Agent supports anywhere else — clone the repo, run the installer, and let it set up its dependencies.
Step 5: Set your environment variables.
Add your model API key (or your Ollama configuration if you’re running models locally on the VPS) to the environment file Hermes Agent reads on startup.
Step 6: Do a test run in the foreground first.
Before you background anything, start Hermes Agent directly in your terminal session and confirm it boots cleanly and can reach your model.
Once that test run behaves, you’re ready to make it permanent.
Keeping Hermes Agent Running 24/7 on Hostinger
A VPS is only useful for an always-on agent if the process survives you closing your SSH session.
The standard way to handle that on Ubuntu is a process manager like pm2, or a systemd service if you want it to restart automatically on server reboot.
Either way, the goal is the same: Hermes Agent keeps running goal loops and Kanban swarms whether or not you’re actively connected.
This is exactly the gap Mike Power was pointing at — local setups die the second your machine sleeps, and a properly daemonised Hostinger install just doesn’t.
🆓 Want the free version of this workflow?
I put a simplified walkthrough of running free AI agents like Hermes inside the FREE AI Money Lab community — no cost to join.
Connecting Hermes Agent to Your Models
Once it’s installed, Hermes Agent still needs a model to actually think with.
You’ve got two realistic options on a Hostinger VPS.
Option one is to point it at a hosted model over an API key, which keeps your VPS plan lightweight since the heavy lifting happens off-server.
Option two is to run MiniMax-M2.7 or Gemma 4 locally through Ollama, directly on the same VPS, which needs more RAM but keeps everything self-contained.
I’ve had good results wiring Codex in alongside Hermes Agent too — I go through that full connection in a separate Codex-to-Hermes guide if you want to add a second model into the same setup.
Which Hostinger VPS Plan Should You Pick?
People ask me this more than anything else, so here’s the honest answer.
If Hermes Agent is only calling out to a hosted model over an API key, you’re not asking the VPS to do any heavy inference, so an entry-level KVM plan with 4GB RAM handles it fine.
If you want MiniMax-M2.7 or Gemma 4 running locally on the box through Ollama, budget for a higher-RAM plan, because the model itself needs to load into memory alongside Hermes Agent’s own processes.
If you’re running Hermes Workspace’s Kanban swarms with several agents active at once, size up again — multiple agents means multiple concurrent processes competing for the same CPU cores.
My rule of thumb: start on the smallest plan that fits your model choice, watch resource usage for a week, and upgrade the plan rather than over-buying from day one.
Hostinger lets you scale a VPS plan up from the same dashboard without rebuilding the server, which is exactly the “scaling is a dashboard click, not a new machine” advantage from the comparison table earlier.
Common Issues When You Install Hermes Agent on Hostinger
A few things trip people up the first time, so here’s what to check before you assume something’s broken.
If the install script fails partway through, it’s almost always a missing dependency — double check Node.js and build tools installed cleanly in step 3 before you re-run it.
If Hermes Agent starts but can’t reach your model, it’s usually the environment file, not the install itself — confirm your API key or Ollama endpoint is actually being read by the process.
If the process dies the moment you close your SSH session, that means you skipped the pm2 or systemd step, so the process was still tied to your terminal session instead of running independently.
If Hermes Workspace’s web interface won’t load from outside the server, check Hostinger’s firewall settings in hPanel — the port it’s running on needs to be explicitly allowed through.
Almost every “it doesn’t work” report I’ve seen traces back to one of those four things, not a genuine bug in Hermes Agent.
Securing Your Hostinger Hermes Agent Setup
Once Hermes Agent is live 24/7, it’s a real production system, not a toy.
Treat it like one.
Keep Hostinger’s automatic backups or snapshots switched on, so a bad update or a corrupted config doesn’t cost you your whole setup.
Lock down SSH access to key-based login rather than password login, and don’t share root credentials across a team when individual accounts will do.
If you’re exposing Hermes Workspace’s web interface publicly, put it behind a subdomain with SSL rather than leaving it open on a bare IP address.
None of this is exotic advice — it’s the same basic hygiene you’d apply to any server that’s doing real work for your business.
Where This Fits With Hermes Desktop and Hermes Workspace
Installing Hermes Agent on Hostinger doesn’t replace Hermes Desktop or Hermes Workspace, it just changes where the agent itself lives.
I’ve written separately about what Hermes Desktop can actually do if you want the offline client walkthrough, and about the Hermes Agent Mission Control dashboard if you want a single screen to watch everything the agent’s doing.
If you’re also weighing up ongoing costs, I broke down what’s actually worth paying for in my best subscription for Hermes Agent post — a Hostinger VPS is usually the cheapest recurring cost in the whole stack.
FAQ: Installing Hermes Agent on Hostinger
Can I install Hermes Agent on Hostinger for free?
Hermes Agent itself is free and open-source, so the only real cost is the Hostinger VPS plan you choose to run it on.
Do I need a powerful VPS to run Hermes Agent on Hostinger?
Not always — if you’re calling a hosted model through an API key a modest plan is fine, but running MiniMax-M2.7 or Gemma 4 locally through Ollama on the server needs more RAM.
Is Hostinger better than running Hermes Agent on my own Mac?
For anyone dealing with fan noise, battery drain, or slowdowns from long agent loops, yes, because a VPS gives Hermes Agent a dedicated environment instead of fighting your laptop for resources.
Can I still use Hermes Desktop if I install Hermes Agent on Hostinger?
Yes, Hermes Desktop and Hermes Workspace both still work as your interface, you’re just changing where the underlying agent process actually runs.
What’s the biggest mistake people make installing Hermes Agent on Hostinger?
Skipping backups and access control — treat a 24/7 agent VPS like the production system it is, snapshots on, SSH locked to keys.
About Julian
I’m Julian Goldie, founder of Goldie Agency, a seven-figure SEO and link-building agency with a team of 70+.
I run the Julian Goldie SEO YouTube channel to 400K+ subscribers, share daily AI and SEO breakdowns to 163K followers on X, and have taught over 29,000 students on Udemy.
I wrote Link Building Mastery, and I run the AI Profit Boardroom, a community of 3,600+ members building real AI agent workflows like this one.
If you’re setting up your own Hermes Agent install on Hostinger and get stuck, that community is where I answer the specific, messy questions that don’t fit neatly into a blog post.
📺 Video notes + links to the tools 👉
