Hermes Pixel Office: Watch Your AI Agents Work

Ever wonder what your AI agents are actually doing?

You hit go, and then you wait.

A blinking cursor, a wall of scrolling text, and no clue whether they are working, stuck, or already finished.

The Hermes Pixel Office fixes that in the most literal way possible: it turns every agent into a tiny pixel character sitting at a desk, and you watch them work.

It is free, it is open source, and it made my whole agent OS more fun to run.

What it is

It is called the Hermes Pixel Office.

It comes from Teknium, one of the people behind Nous Research — the same team that builds Hermes Agent.

It is free and open source. Nothing to sign up for, nothing to unlock.

The problem it solves

Hermes agents normally work in the background.

They run tools. They spin up helper agents. They get real work done.

And you cannot see any of it.

You sit there staring at scrolling text, hoping it is going fine and not quietly stuck on something.

That is a genuinely bad way to run a business process, and most people have just accepted it.

How it actually works

The Pixel Office takes all of that invisible activity and turns it into a little cartoon office.

Every agent becomes a pixel character sitting at a desk.

You watch them walk in through the door, sit down, and get to work.

When they finish, they stand up and walk right back out.

That is it.

You read the whole thing at a glance, like looking through a window into a real room.

This matters more for me than it might for you, because my entire agent OS runs on Hermes.

Hermes is the brain behind the whole thing.

So when I open the Pixel Office, I am literally watching my agent OS agents work — every single one of them, live, in one little room on my screen.

Two ways to run it

You can open it in your browser as its own page.

Or you can drop it straight into VS Code, so the office sits right next to your code while you work.

Same office either way. Your pick.

How to read the characters

Each agent session is one character.

It does not matter where the session came from — the command line, a chat app, or a job you set to run on a schedule.

They all show up as their own little person walking in.

The helper agents, the ones a bigger agent spins up to handle a side task, get a gold collar and a name tag showing their job.

So the second you look, you know who is the main worker and who is a helper it pulled in.

They act out what they are doing

This is the part that makes it click.

When an agent is writing a file, its character is typing.

When it is reading something, it has a little book open.

When it is browsing the web, it is browsing.

When it is running something in the terminal, you see a green monitor flicker.

You do not have to read a single log line to know what is going on.

You just look at the room.

🔥 Want this exact setup on your own machine?

Inside the AI Profit Boardroom you get the full Agent OS zip ready to install — the same setup I run with Hermes as the brain, so the Pixel Office plugs straight in. Plus a 30-day roadmap, step-by-step Hermes walkthroughs, and four live coaching calls every week with 4,000+ members.

→ Get access here

Approval flags: the genuinely useful bit

Sometimes an agent hits a command that could be risky, and it needs you to say yes or no before it goes ahead.

In the Pixel Office, that agent throws up a bright red exclamation mark over its head.

The top of the screen shows a little counter — something like "2 waiting".

So instead of scrolling through walls of text trying to work out whether anything is stuck waiting on you, you glance over and see exactly who needs you.

There is even an optional chime you can switch on.

It dings when an agent needs your okay, or when a helper finishes its job.

It is off by default. Flip it on and you do not even have to be looking at the screen.

One shared office

Every Hermes agent running on your machine shares the same office.

So if you have a few different things going at once, they all pile into one room together.

One window. Your whole team.

The biggest myth about it

When people see a live dashboard like this, they assume it slows their agents down, or gets in the way, or drags on them under the hood.

It does not.

The Pixel Office is a window and nothing more.

It watches what your agents are already doing and draws it out for you.

It never blocks them, never changes their work, and never adds any weight to how they run.

Your agents do not even know it is there.

So you get the whole show without giving anything up, and that is what makes it click for me.

I am not trading performance for visibility. I just get to see.

Old way vs new way

Old way	New way
Stare at scrolling logs and hope.	Glance at a room and know.
Find out something stalled an hour later.	See the character stop moving immediately.
Hunt through text for approval prompts.	A red flag over the agent's head and a counter at the top.
No idea which agent is which.	Gold collars and name tags for helper agents.
Guess whether a batch finished.	Characters stand up and walk out when they are done.

Three ways I actually use it

Watching a content series get built. I kicked off a batch of Hermes agents inside my agent OS to draft hooks, scripts and captions. In the Pixel Office I watched each helper pop up as a gold-collared character — one typing away, one reading through notes, all working at the same time. When one hit a step that needed my okay, the red flag appeared over its head, so I knew the exact second to step in.

Checking a member onboarding flow. I had agents lay out welcome messages, first-call prep and resource walkthroughs. I watched each character finish its task and walk out the door, which told me at one glance that the whole flow was done and nothing was silently stuck.

Running a lead generation system. Fire off a group of agents to plan the content, the hooks and the follow-ups, then watch the office fill up. At a glance I know which are still working, which finished and left, and which one is waiting on me. That is the whole thing from one small screen instead of ten open tabs.

It does not do the work for me.

It finally lets me see the work.

Four pro tips

Try the demo first. It ships with a demo mode that fills the office with pretend agents, so you can learn what every character means before you connect your real setup.

Put it in VS Code if you live there. You get the same office in a panel next to your code, plus a "+ agent" button that starts a fresh Hermes run without leaving your editor.

Remember it is a window, not a controller. Keep your agents' own safety rules turned on. The red flag only tells you when to look — you are still the one who says yes or no to anything risky.

It works over a remote connection out of the box. If you run your agent OS on a server somewhere, like I do, you still get the office on your own screen with no extra wrestling.

Who this is for

If you run one agent occasionally, you probably do not need it.

If you run batches of agents with helper agents underneath them, it turns an unreadable wall of text into something you can supervise in a second.

And if your whole operating system runs on Hermes, it is the missing window into your own business.

FAQ

What is the Hermes Pixel Office? A free, open source visual dashboard from Teknium at Nous Research that shows every Hermes agent as a pixel character working in a shared cartoon office.

Does it slow my agents down? No. It is a window, not a controller — it only draws what your agents are already doing, and it never blocks or changes their work.

How do I tell helper agents apart? Helper agents get a gold collar and a name tag showing their job, so main workers and helpers are obvious at a glance.

How do I know when an agent needs approval? A bright red exclamation mark appears over that character's head, and a counter at the top shows how many are waiting. There is an optional chime too.

Can I run it in VS Code? Yes — either as its own browser page or as a panel inside VS Code, with a "+ agent" button to start a new Hermes run without leaving the editor.

About Julian

I'm Julian Goldie — AI entrepreneur, SEO expert, and founder of the AI Profit Boardroom (4,000+ members). I help business owners scale with AI agents, automation, and SEO.