agentos.guide › blog
The Best Hermes Agent Setup, And How It Becomes Your Agent OS
The best Hermes Agent setup is Hermes Desktop with a Nous Portal model as the main brain, a local Ollama model as the fallback, a plain-text memory vault and one dashboard on top.
I call it the three-layer setup, because it gives the agent a brain, a cockpit and a memory.
The free setup is my second pick, and the one-click beginner setup is my third.
Here is the ranked list first, then a seven-step quick-start and how each piece fits an Agent OS.
The best Hermes Agent setups, ranked
Here is the full ranking, with who each setup suits and where it fits in an Agent OS.
| Rank | Setup | Best for | Cost | Where it sits in an Agent OS |
|---|---|---|---|---|
| 1 | Three-layer setup | Most people | Free software, with a free or paid brain | It is the whole OS, with a brain, a cockpit and a memory. |
| 2 | Free setup | Testing and light daily jobs | $0 | It is the brain layer on a free model. |
| 3 | One-click beginner setup | First-time users | Free app | It is the brain layer with Hermes Desktop as the cockpit. |
| 4 | Profile-per-model setup | Power users | Depends on your models | Each profile becomes a chip in the cockpit. |
| 5 | Always-on server setup | Businesses and teams | A server or Hermes Cloud, plus the model | It is a remote brain you reach from chat apps. |
| 6 | Local private setup | Private data | $0, plus hardware | It is the offline fallback brain. |
This is my own first-person ranking, and it comes from what I run every day and what I have covered in my guides.
It is an operator's opinion and not a measured benchmark.
I ranked on four things, which are speed to a working agent, monthly cost, memory between sessions and whether it works while I am away.
What a Hermes Agent setup actually is
Hermes Agent is a free, open-source AI agent from Nous Research that can use tools, run commands and remember things.
It does not come with one fixed model, so you choose the brain.
A setup is five choices, which are the install, the front end, the main model, the memory and the plugins.
Most people only make the first choice, and that is why their agent ends up abandoned after a weekend.
The agent is not weak.
It simply has no dashboard to drive it from, no memory of yesterday and no job running overnight.
Quick-start: build the best Hermes Agent setup in seven steps
This is the order I would follow on a fresh machine.
| Step | What you run or do | What you get |
|---|---|---|
| 1 | Install Hermes Desktop, or run curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash. | Hermes Agent is on your machine. |
| 2 | Run hermes setup --portal. | You are logged in to Nous Portal with a model picked. |
| 3 | Run hermes fallback add and choose a second model. | A rate limit no longer stops your work. |
| 4 | Make a notes folder with an about-me note, a decisions log and a daily log. | The agent has a memory it can read and write. |
| 5 | Run hermes profile create research and hermes profile create builder. | You have two agents with separate memory and rules. |
| 6 | Run hermes dashboard. | A web dashboard opens on your own machine at port 9119. |
| 7 | Run hermes plugins search, then install one plugin. | You add one new power and can test it properly. |
The Mac desktop package supports Apple Silicon only, and Windows uses the App Installer package.
On Windows without WSL, the official docs give a PowerShell install, which is iex (irm https://hermes-agent.nousresearch.com/install.ps1).
I checked every command in this table against hermes --help on version 0.21.4.
Do the memory step before the dashboard step, because a remembered business is worth more than a pretty screen.
🔥 Want the exact Hermes Agent setup in the Agent OS setup I run inside my Agent OS?
Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.
How this setup fits an Agent OS workflow
An Agent OS is the wrapper that turns one agent into a system you run a business from.
My 3-Layer Hermes OS guide splits that wrapper into a brain, a cockpit and a memory.
| Layer | What it is | Hermes commands that build it |
|---|---|---|
| The brain | Hermes plus the models it thinks with | hermes setup, hermes model, hermes fallback add |
| The cockpit | One place where every job has a button | hermes dashboard, hermes desktop, hermes profile create, hermes kanban |
| The memory | Notes the agent reads before work and writes after | hermes memory status, hermes memory setup, plus your notes vault |
In my own Agent OS, every profile in the Hermes profiles folder shows up as a chip in the chat automatically.
Each chip is one employee persona, with its own model, memory and rules.
My Hermes Command Desk guide shows the same agent behind thirteen different tabs, from chat to voice to a kanban board.
Bigger jobs become tickets on that board rather than prompts, and a Hermes worker picks each one up.
Every exchange is saved to my Obsidian vault, which held 1,018 sessions when I wrote that guide.
You do not need thirteen tabs to start.
You need a chat surface, two profiles and a folder where the outputs land.
Setup 1: the three-layer setup
This is my top pick, because it is the only setup that fixes memory, control and cost at once.
The brain is a Nous Portal model with a local Ollama model as the fallback.
The cockpit is Hermes Desktop or the web dashboard.
The memory is a plain-text notes folder, and I use Obsidian for mine.
Three notes get you started, which are an about-me note, a decisions log and a daily log.
Tell Hermes to read the folder at the start of each session, and end each session by asking it to log what it did.
Hermes also has built-in memory that is always active, and hermes memory status shows what is switched on.
Setup 2: the free setup
Hermes is free, so the model is the only layer that can send you a bill.
Run hermes model, choose Nous Portal and pick a model that is marked free.
Some models appear in both a free and a paid version, so check the label before you confirm.
The other free route is ollama launch hermes with a free Ollama cloud model, which comes with token limits.
Free models get rate limited at busy times, so add a second one with hermes fallback add.
My Solar Mini 4 and Hermes guide shows a free Portal model answering in a real session.
Setup 3: the one-click beginner setup
Download the Hermes Desktop package from the Hermes website and install it like any app.
Sign in to Nous Portal inside the app, pick a model and type a test message.
The bundled package already contains the agent and its dependencies, so there is nothing to build on first launch.
This is the right place to start if you have never opened a terminal.
My best Hermes Agent UI ranking compares Desktop with the other front ends.
Setup 4: the profile-per-model setup
A profile is a separate Hermes home with its own config, keys, memory, sessions and skills.
Run hermes profile create coder, then run hermes -p coder to chat with it.
Every profile also becomes its own command, so coder chat works too.
I keep one profile per model, which makes switching brains a single flag.
To run Claude on a subscription you already pay for, run hermes plugins install claude-subscription-directsdk.
Then set the provider to claude-subscription-directsdk-experimental and the default model to opus in that profile's config file.
That plugin needs Hermes 0.21.4 or newer and a Claude Pro or Max plan, and Nous marks it experimental.
The full walkthrough is in my Hermes Claude subscription guide.
The official docs warn that two running agents should never share one profile, because both write memory.
My best Hermes Agent model ranking helps you decide which brain each profile gets.
🔥 Want the exact Hermes Agent setup in the Agent OS setup I run inside my Agent OS?
Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.
Setup 5: the always-on server setup
A laptop agent stops when the lid closes, so this setup moves Hermes somewhere that never sleeps.
Connect to a Linux server over SSH and run the same install script, then run hermes setup.
Run hermes gateway setup to connect a chat platform, then hermes gateway install and hermes gateway start.
Check it with hermes gateway status before you rely on it.
The dashboard binds to 127.0.0.1 by default, so reach it through an SSH tunnel or a private network such as Tailscale.
There is also an official Docker image, which is nousresearch/hermes-agent.
Hermes Cloud on Nous Portal is the managed route, and my Hermes Cloud guide covers it.
When I wrote that guide, a Cloud Agent needed $10 in credits or an active Nous subscription to deploy, and it scaled to zero when idle.
My best Hermes Agent VPS post covers the server choice.
Setup 6: the local private setup
If your data cannot leave your machine, run the brain locally through Ollama.
On an 8GB machine I use LFM2.5 2.6B from LiquidAI, and the full commands are in my run Hermes free forever guide.
Pull it with ollama pull hf.co/LiquidAI/LFM2.5-2.6B-GGUF:Q4_K_M, because the short tag gave me a different model.
I also had to rebuild it with Ollama's native LFM renderer, because the direct download broke agent loops.
It ran at about 140 tokens per second on my Mac and handled five-step tool chains cleanly.
It also failed at building a full web app, so I use it for errands and keep a bigger brain one profile away.
The commands worth keeping next to your keyboard
These cover most of what you will do after the first week.
| Command | What it does |
|---|---|
hermes setup | It runs the interactive setup wizard. |
hermes model | It picks your default provider and model. |
hermes model --refresh | It clears the cached model list and fetches a fresh one. |
hermes profile list | It lists every profile you have made. |
hermes gateway setup | It connects messaging platforms such as Telegram and Slack. |
hermes cron create | It creates a scheduled job. |
hermes doctor | It checks your configuration and dependencies. |
hermes backup | It zips your Hermes home so you can restore it. |
hermes update | It updates Hermes to the latest version. |
There is also an emergency stop, which is hermes pause, and it halts scheduled jobs and new gateway turns.
Mistakes that break a Hermes Agent setup
The first mistake is skipping the memory layer, which leaves you re-explaining your business every morning.
The second mistake is installing lots of plugins at once, when one at a time is the only way to find the one that breaks things.
The third mistake is running without a fallback model, because a rate limit will then stop your whole day.
The fourth mistake is opening the dashboard to the public internet instead of tunnelling in.
The fifth mistake is updating without running hermes backup first.
The sixth mistake is chasing every new model, when the point of an Agent OS is that a new model is just a dropdown change.
What I would do in your first week
On day one, install Hermes and get one reply from a free model.
On day two, make the notes folder and write the about-me note.
On day three, create two profiles, one for research and one for building.
For the rest of the week, turn every request you type twice into a profile or a button.
After that, add the always-on gateway only when a job genuinely needs to run without you.
Also On Our Network
FAQ
What is the best Hermes Agent setup?
My pick is the three-layer setup: Hermes Desktop or the official install script, a Nous Portal model as the main brain, a local Ollama fallback, a plain-text memory vault and a dashboard. It is my own ranking from daily use, not a measured benchmark.
How does a Hermes Agent setup become an Agent OS?
An Agent OS is the wrapper around the agent. Add a cockpit where every job has a button and a memory vault the agent reads and writes, and Hermes stops being a chat window and becomes a system you can run a business from.
What command installs Hermes Agent?
On Linux, macOS or WSL2, run curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash and then hermes setup. On macOS and Windows you can download the Hermes Desktop package instead.
Do I need to pay for a model to use Hermes Agent?
No. You can pick a model marked free on Nous Portal, use a free Ollama cloud model with token limits, or run a local model on your own machine. Free models get rate limited, so add a fallback.
How many Hermes profiles should I create?
Start with two, one for research and one for building. Add a profile whenever a job needs its own model, memory or rules, and never run two agents on the same profile at once.
Can Hermes Agent run when my laptop is closed?
Yes, if you install it on a server and run the messaging gateway as a background service with hermes gateway install and hermes gateway start, or deploy a managed agent on Hermes Cloud.
Final word on the best Hermes Agent setup
Install Hermes, plug in a brain, then add the memory before you add anything else.
Grow the cockpit one button at a time, and you will end up with the best Hermes Agent setup.
About Julian
I'm Julian Goldie, an AI entrepreneur, SEO expert and the founder of the AI Profit Boardroom, which has 3,400+ members.
I help business owners scale with AI agents, automation and SEO.
My YouTube channel has 400,000+ subscribers, and I run Goldie Agency, a seven-figure SEO agency.
Related guides on agentos.guide
→ Claude Opus Inside Hermes On Your Subscription
Get the whole Agent OS
The installable system behind every guide on this site — dashboard, agents, memory and pipelines — updated as new tools land, with 3,400+ members and four coaching calls a week.
Join the AI Profit Boardroom →📺 Video notes + links to the tools 👉
🎥 Learn how I make these videos 👉