Hermes Agent, installed. And locked down.
Hermes Agent is an open-source AI agent from Nous Research that you run yourself. We install it on hardware you own, limit what it can reach, and set it up for real work.
01 · In plain English
Hermes Agent is a self-hosted AI agent. It runs on a server or computer you control and works through tasks using the tools you connect it to.
- You run it. On a machine you own or rent, powered by an AI model you choose.
- It has real access. Whatever you connect it to. That's the point, and the risk.
- Security is the job. Like any agent, it can be misled by instructions hidden in emails or web pages. We set it up to limit the damage.
02 · What you get
One fixed price for teams of up to 20 people (bigger team? ask us). Here's everything that's included.
Where it runs, chosen with you
A small cloud server or a computer at your office. We tell you the running costs before you commit.
Installed and hardened
Its own accounts, no admin rights, secrets stored safely, remote access locked down.
Model chosen with you
Which AI model powers it, balancing cost, quality and where your data goes.
Limits on what it can touch
Tools connected one at a time, read-only first, with a person approving anything it sends or changes.
Your first tasks
Two or three real jobs set up and tested with you.
Your AI policy
A short, plain-English policy written with you: what the tool may touch, what must never go in, and who checks its work.
A one-page handover
Who your admin is, how to add and remove people, what's connected, and how to switch it off. Everything you need to run it without us.
03 · Who pays what
| What | Paid to | How much |
|---|---|---|
| AI Setup | AI Workers | $1,500 once, fixed |
| Server or computer | Your hosting provider, directly | A small monthly cloud bill, or hardware you already own |
| AI model usage | The model provider, directly | Pay as you go; we estimate it with you |
| Ongoing support (optional) | AI Workers | Only if you want it; ask us |
04 · Before you start
- Prompt injection is real. Limits and human approval are how you contain it.
- Someone has to maintain it. Updates matter. We show you how, or can do it as ongoing support.
- It might not be the right tool. For most teams, a properly set-up Claude or ChatGPT is simpler. We'll tell you honestly.
Not sure which tool is right? Read our ChatGPT vs Claude vs Copilot guide, see what goes in an AI policy, or read what an AI agent is.
05 · Other tools we set up
- Claude: Claude on a business plan, connected to the tools you approve, with Cowork and Claude Code if you need them.
- ChatGPT: Your team off personal ChatGPT accounts and onto a business plan you own, with GPTs built for your work.
- Microsoft Copilot: Copilot in Microsoft 365, with your sharing permissions checked first so it can't surface the wrong files.
- Grok Bot: An AI agent that works through tasks for your team. We set the limits before it starts.
- OpenClaw: The open-source agent, installed on hardware you own and locked down so it only does what you allow.
Questions
What is Hermes Agent?
An open-source AI agent from Nous Research that you host yourself and connect to the tools you choose.
How much does Hermes Agent setup cost?
$1,500, fixed, paid to us. You also pay for the server (or use one you have) and the AI model, directly to those providers.
Hermes Agent or OpenClaw?
Both are self-hosted agents. The right one depends on the tasks, the chat apps you use and who'll maintain it. We'll help you choose.
Is it safe for business?
With limits, yes. We give it its own accounts, the least access it needs, and a person approving anything that leaves the business.
Next step
Tell us how many people are on your team and what you want Hermes Agent to do. We'll tell you the safest way to set it up.