agentos.guide › blog
The Hermes Agent Robot: The Autonomous Lamp Quick Start, And How It Fits Your Agent OS
The Hermes agent robot is the Autonomous Lamp, a fully-assembled open-source desk robot from Autonomous that can run Hermes Agent as its brain through lampOS.
It has a camera, microphones, speakers, a light ring and a five-servo arm, and the Lamp's device file in the repo sets Hermes as its default agent.
On 24 October 2026, Autonomous's official page listed it at $999 for one unit, with free shipping.
This page is the quick start, plus how an agent with a body fits an Agent OS workflow.
The Hermes agent robot at a glance
This table is the whole product on one screen.
| Question | What Autonomous's page and the lampOS README say |
|---|---|
| What is it? | It is the Autonomous Lamp, a fully-assembled open-source companion robot shaped like a desk lamp. |
| Who makes it? | Autonomous makes the Lamp and its operating system, lampOS. |
| What does it cost? | The official page listed $999 for one unit with free shipping on 24 October 2026. |
| Which agents can drive it? | The README lists Hermes, Claude Code, Codex, OpenClaw, OpenCode and PicoClaw. |
| Which agent is the default? | The Lamp's ROBOT.md sets Hermes as the default gateway. |
| What do I need to start? | Autonomous's FAQ says you need Wi-Fi and the Autonomous app. |
| Is the OS open source? | Yes, it is Apache-2.0 outside hal/, and hal/ is GPL-3.0. |
The repo is public at github.com/autonomous-ai/lampOS, and it had about 490 stars on 24 October 2026.
I covered the Lamp in my video using Autonomous's own footage, but I do not own one and I have not tested it, so this guide is built from Autonomous's page and the lampOS README.
What the Hermes agent robot is
The Hermes agent robot is what people call the Autonomous Lamp when it runs Hermes Agent as its brain.
Hermes Agent is Nous Research's open-source agent, and the Lamp gives it eyes, ears, a voice and an arm it can move.
Autonomous's announcement on X, which I showed in my video, said "We just gave Hermes a body, one eye, and two ears."
The README calls the software "Physical AI Operating System" and says Autonomous is building the "Android" for physical AI agents.
The OS runs on the robot and coordinates the hardware and the agent's tasks.
The agent never drives a motor directly, because hardware skills emit markers like [HW:/path:{json}] that the OS checks and runs.
The same OS also runs on Autonomous Intern and on Hugging Face's Reachy Mini, so skills are not tied to one product.
The Hermes agent robot quick start
These are the six steps from the README, in its own order.
| Step | What you do | What the README says happens |
|---|---|---|
| 1 | Tap Add robot and then Lamp in the Autonomous app. | The app starts adding your Lamp. |
| 2 | Pick your Wi-Fi network in the app. | The app joins the robot's hotspot and hands over the keys and pairing. |
| 3 | Say something to it. | It turns to look at you, the ring lights up and it answers. |
| 4 | Install a skill from the Skill Store. | It takes one tap and is live on the next conversation. |
| 5 | Type what you want it to do in the app. | The app writes the skill for you. |
| 6 | Edit SOUL.md. | The Lamp is someone else on the next turn. |
The app is on iOS and Android, and both links come from the README.
The Lamp's ROBOT.md turns on a wake word by default, listening for "hey lamp", "hey autonomous" or "hey" plus your agent's name.
Autonomous's FAQ adds that you can also use the attention control set up on your Lamp.
Simple requests such as "Turn off the lights" or "Make this lamp violet" are handled by local rules first, and anything uncertain goes to the main agent.
🔥 Want the exact Hermes Agent 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.
What the Lamp gives your agent to work with
This is the hardware your agent can use, from the official spec sheet.
| Part | What the official page lists |
|---|---|
| Movement | It uses 5 position-feedback servos. |
| Vision | It has a wide-angle camera with face tracking and motion sensing. |
| Audio | It has a built-in microphone array and built-in stereo speakers. |
| Compute | It runs on an 8-core ARM64 CPU with 6 GB of RAM. |
| Connectivity | It connects over Wi-Fi 6. |
| Power and size | It takes 12V 5A DC and measures 7.87 × 7.87 × 18.43 inches. |
| Weight and warranty | It weighs 6.6 lbs and has a 1-year warranty. |
| Room sensing | It lists CO₂, particle, temperature and humidity sensors. |
The page lists CO₂ accuracy at ±100 ppm and typical temperature accuracy at 0.45 °C.
One line of the page refers to "Lamp Pro's room sensors", so check which model you order if room sensing matters.
The README says the hardware layer exposes 111 HTTP endpoints, with live Swagger docs on the robot.
What the Hermes agent robot costs
On 24 October 2026, the official Lamp page listed $999 for one unit, with free shipping.
The page also shows lower per-unit prices for three or more Lamps.
The other cost is the brain, because the model calls run on Autonomous's hosted AI by default, or on your own keys if you switch.
I am not quoting model costs, because they depend on which provider and model you pick.
How to make Hermes the brain
The README says you run Hermes, Claude Code, Codex, OpenClaw, OpenCode or PicoClaw, then choose a compatible model and configure its voice, personality and tools.
All six sit behind one gateway, and you can switch live from the web UI while the persona, memory and connectors move with it.
| Where the brain runs | What Autonomous says |
|---|---|
| Autonomous's hosted AI | The FAQ says the Lamp runs on hosted AI right after setup, including the brain, the voice, face recognition and mood. |
| Your own agent or keys | The FAQ says you can switch to your own agent or keys after setup. |
| A cloud model you choose | The page says you can connect a cloud model from a provider you pick. |
| A local model | The FAQ says yes with a compatible local model endpoint. |
| Fully offline | The FAQ says you also need compatible speech and vision services on your own hardware. |
The README says model inference may use remote services, depending on your configuration.
If you want the Hermes side in more depth, my guide to Hermes Agent v0.22 covers the current release.
Skills, soul and safety on the Hermes agent robot
The README says any robot can be described in four markdown files.
ROBOT.md is the body, SOUL.md is the self, SAFETY.md is the bounds and SKILL.md is one thing it can do.
A skill is one folder with one SKILL.md inside, and skills declare the capabilities they need, so they only install on compatible robots.
The README lists a harness-use skill that sends coding and research tasks to agents running in Autonomous's Harness app on a paired computer.
It also lists a computer-use skill that lets the Lamp's agent use Mac apps through a paired Autonomous Buddy, which bundles Cua Driver.
If computer-use agents are new to you, my guide to Hermes CUA explains how they work.
The safety gate is "a pure function" of SAFETY.md, sitting below the agent with no model in the loop.
| Limit | What the Lamp's SAFETY.md sets |
|---|---|
| Brightness | It caps the daytime light at 120 on a 0 to 255 scale. |
| Quiet hours | It dims the ring to a ceiling of 40 from 22:00 to 07:00. |
| Motion speed | It caps every joint at 120 degrees per second. |
| Stop | It keeps stop and release always available, outside the agent's control. |
The README also links to a safety doc that lists what the gate does not cover yet.
🔥 Want the exact Hermes Agent 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 the Hermes agent robot fits an Agent OS workflow
My Agent OS runs a crew of agents, including Hermes, under one shared memory, so the question I care about is where a body fits in that crew.
I have not connected a Lamp to my own Agent OS, so this section maps the ideas rather than describing a tested setup.
| Agent OS idea | How the Lamp maps to it | Why it matters |
|---|---|---|
| One brain, many bodies | The README says six agents sit behind one gateway and switch live. | You can keep the agent you trust and change only the body. |
| Skills as files | Each skill is one folder with one SKILL.md. | That matches how Hermes skills already work on a computer. |
| Persona as a file | SOUL.md sets who it is and how it talks. | You can give a front-desk agent a different character from a research agent. |
| Hand-offs to a computer | The harness-use skill sends tasks to agents in Autonomous's Harness app on a paired computer. | The Lamp becomes a voice front end for work that runs elsewhere. |
| Safety outside the prompt | SAFETY.md is enforced with no model in the loop. | No agent in the crew can talk its way past the limits. |
The way I see it, the Lamp is a front desk for your agents.
The heavy work still runs on your computer or in the cloud, and the Lamp is the part that listens, answers out loud and moves.
That is why harness-use is the skill I would study first, because it is Autonomous's own route from the robot to agents on another machine.
If you want to talk through where an agent with a body could fit your business, you can book a free SEO strategy session with my team.
The Hermes agent robot limits to plan around
These limits come from Autonomous's own FAQ and the README.
| Limit | What I would do about it |
|---|---|
| The default brain is Autonomous's hosted AI. | I would decide early whether to keep it or switch to my own agent and keys. |
| A local model alone does not make the Lamp offline. | I would plan local speech and vision services if privacy is the main goal. |
| Custom behaviours may need code and testing. | I would start with Skill Store skills and the in-app skill writer. |
| One line of the page calls the air sensors Lamp Pro's room sensors. | I would confirm the model before buying for room monitoring. |
| Face or voice recognition needs enrolment and compatible services. | I would check what my setup supports before relying on it. |
Autonomous's FAQ is honest that custom behaviours, app connections and experiences for other people may need code and testing.
I would add one more of my own, which is to test everything in the real room before you rely on it.
My three takeaways on the Hermes agent robot
First, agents are leaving the screen, because the same Hermes you run in a terminal can now answer out loud and move.
Second, the safety design is right, because the limits live in a file the OS enforces and not in a prompt.
Third, the open OS is the real story, because a skill you write for the Lamp can work on other robots too.
Also On Our Network
FAQ
What is the Hermes agent robot?
It is the Autonomous Lamp, an open-source desk robot from Autonomous that can run Hermes Agent as its brain through lampOS. Hermes is one of six supported agents and is the default in the Lamp's device file.
How much does the Autonomous Lamp cost?
Autonomous's official Lamp page listed it at $999 for one unit with free shipping on 24 October 2026.
How do I set up the Hermes agent robot?
The README gives six steps: add the robot in the Autonomous app, set up Wi-Fi, talk to it, install a skill from the Skill Store, type what you want to build your own skill, and edit SOUL.md.
Which agents can run on the Autonomous Lamp?
The README lists Hermes, Claude Code, Codex, OpenClaw, OpenCode and PicoClaw, and you can switch between them live from the web UI.
Can the Hermes agent robot run fully offline?
Autonomous says a compatible local model works, but fully local conversation also needs compatible speech and vision services on your own hardware.
How does the Lamp keep its movement safe?
The OS enforces SAFETY.md below the agent with no model in the loop, capping brightness, quiet-hours light, joint speed and keeping stop always available.
Final word on the Hermes agent robot
Set it up on hosted AI first, switch the brain to Hermes, install one skill and read the safety file before you build anything of your own.
If you want a second opinion on which agents fit your business, you can book a free SEO strategy session with my team.
If you want the whole Agent OS with Hermes profiles, skills and training already set up, it lives inside the AI Profit Boardroom, with 3,400+ members and four coaching calls a week.
That is the fastest way to get real work out of the Hermes agent robot.
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
→ Hermes Agent v0.22: The Quick Start
→ The Hermes Computer Use Skill
→ The Best Hermes Agent Use Cases
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 👉