Search "make money with Hermes Agent" and you will find trading bots and five-figure screenshots. Read the actual user stories and the pattern is duller: someone automates the repetitive parts of work they already do, or they charge to set that up for other people. That is where the money is, and it is worth being precise about it.
01. The honest shape of it
two models, both boring, both real
Almost every documented case is one of two things. Either you use Hermes to automate your own grunt work, summaries, monitoring, first drafts, ops, so you spend your time on the part that pays. Or you sell the setup: you install and tune Hermes for a business that will not do it themselves. The second one is a service business, and it is the clearer path to revenue.
The value is not the install. It is adapting the thing to how a business actually works.
02. The install-and-adapt model
a real number, from a real person
A Reddit user posting as pacmanpill runs a service installing and adapting Hermes for small businesses in France, and reported roughly 2,700 euros in the most recent month. Their own framing is the important bit: "the value is adapting it to the company's real workflows," not running the install script.
That is the offer. A small business will not read the docs, wire a Telegram bot, write context files, and maintain a skills folder. Someone who has done it ten times can stand that up in a day and charge for the judgment, not the typing. Another builder, going by NathanWilbanks, is on day 297 of a daily automation streak and puts the client-work value automated over that run past 100,000 dollars.
03. Business ops, wired to chat
the same pattern at different sizes
| Who | What they wired up |
|---|---|
| scotty529, at Gumroad | support tickets, mention monitoring, finance-doc updates, on a dedicated Mac running cron jobs |
| Miles Deutscher | the whole business, notes, SOPs, meeting logs, connected to Hermes alongside an Obsidian vault |
| quxiaoyin | three agents split by job: blog posts, competitor research, news monitoring, code review |
| Godzillaton | 11 WhatsApp construction groups aggregated, 82 messages down to 3 lines |
None of these are moonshots. They are a person taking the reading, triage, and drafting they used to do by hand and handing it to an agent that reports back in a chat app.
04. Lead generation and outreach
the one that looks most like a service
One documented job had Hermes scrape a city for a specific kind of business without a website, then write a tailored pitch angle for each one. It ran on a server costing 24 cents an hour. Another builder wired Hermes to an email-lookup tool through an MCP connection for outreach.
This is the clearest "sellable" use. A business wants qualified leads with a reason-to-call attached. An agent that can search, filter, and draft does that overnight for the price of the compute. The judgment you charge for is knowing which businesses are worth contacting and what actually makes them reply.
05. What the economics look like
cheap to run, that is the point
A user going by SuperALfun runs a full stack, a small Hetzner VPS, Ansible, Docker containers, for around nine euros a month, and it delivers a daily summary of about fifty dividend funds to Telegram. The lead-gen job above ran at 24 cents an hour. Production users mostly run mid-size models, not frontier ones, because most of this work is grunt work where speed beats brilliance.
So the shape of a service is: near-zero infrastructure cost, model tokens that scale with use, and your time. The margin is good precisely because the running cost is so low.
06. The claims to be skeptical of
the internet's favourite Hermes stories
There are posts about Hermes trading bots turning 100 dollars into 216 in 48 hours on weather derivatives, and Polymarket bots with four-layer analysis pipelines. Treat these exactly the way you treat every trading claim online: as a story, not a method. Survivorship bias is brutal in that genre, the losing runs do not get posted, and "an agent placed the trades" does not make the edge real.
If a Hermes use case sounds like passive income, it is probably a trade with extra steps. The repeatable money is in doing known work faster, or selling that setup to someone who will not do it themselves.
07. Could you run this
a quick self-check
- Have you set up Hermes end to end at least twice, including a chat bot and a cron job?
- Can you name three repetitive tasks a specific local business does that an agent could take?
- Are you comfortable maintaining someone else's skills folder and context files after handover?
- Can you explain to a non-technical owner what the agent will and will not touch?
If those are yes, the install-and-adapt service is a real, low-overhead thing to offer. If they are not yet, the setup guide and skill-loop guide are where you close the gap.
Part of the Hermes set
This is the last guide in the set. The overview frames it, setup gets you running, the others go deep on one piece each.
- Overview
- Setup and first run
- The skill loop
- Integrations and automations
- Where it pays off
08. If you want help building one
what I actually do
I build automation and agent systems for businesses across AI, software, and operations, and I run workshops for teams that want to do it in-house. If you are trying to turn a Hermes setup into a service, or you want an agent wired into your own ops and are not sure where the risk is, that is the kind of thing a call is for.
No pitch on this page beyond that line. Book a call if it is useful, or take the guides and do it yourself. Both are fine.
Sources and further reading
- The official user stories. Every name and number in this piece is quoted there, each with a link to the original post.
- Miles Deutscher on X, one of the ops cases quoted above.
- The repo and awesome-hermes-agent for the deployment recipes and MCP integrations referenced.
- To build the setup itself: the setup guide and the skill-loop guide.
// Free newsletter
I send out guides like this every week
Real setups, real sources, no hype. Drop your email and I'll send you the next one.