White-Label Hermes Agent Hosting

The short answer

White-label Hermes hosting lets you package Hermes Agent as your own product. On OpenClaw Launch you either publish a ready-made storefront under your own name, or create a bot per client through a REST API. Each bot costs $0.007 an hour on Lite, about $5 for a 30-day month, billed from a prepaid balance with no deploy fee. You set the retail price, you own the client relationship, and we run the infrastructure out of sight.

Who this is for

Agencies with a client list, consultants whose customers keep asking for the same assistant, and SaaS founders who want to put an always-on agent inside their product. The common thread: you understand a market well enough to sell an assistant into it, but you do not want to run containers, patch servers, or get woken up when a client's bot goes quiet at 3am.

Hermes Agent is a good fit for this because it is light and starts quickly, which matters when you are deploying one bot per client, and because it builds its own skills from the work it does, so a client's assistant tends to get more useful the longer they keep it.

Two ways to sell Hermes under your brand

The storefront runs inside one of your own bots on a regular OpenClaw Launch plan, so you can see a working shop before you write anything. The API route looks like this:

BASE=https://openclawlaunch.com/api/v1

# One Hermes bot for one client, with its own assistant name
curl -X POST $BASE/instances \
  -H "Authorization: Bearer $RESELL_KEY" \
  -H "Content-Type: application/json" \
  -d '{"framework":"hermes","assistant_name":"Ava"}'

# Send it a message and stream the reply
curl -N -X POST $BASE/responses \
  -H "Authorization: Bearer $RESELL_KEY" \
  -H "Content-Type: application/json" \
  -d '{"instance_id":"<id>","input":"Hello","session_id":"client-42"}'

Keep the session_id stable per client or per thread so the bot remembers the conversation. An assistant name cannot contain “Hermes” or “OpenClaw”, which suits a white-label product anyway. A new bot is usually ready in seconds; when no pre-warmed container is free it can take a minute or more, so let your sign-up flow wait for it.

What white-label Hermes hosting costs

Billing is prepaid and hourly, so a test bot you create and delete after an hour costs about a cent. The part that surprises people: the meter runs until a bot is deleted, not until it is stopped, because a stopped bot still holds its storage and its slot on a server. Optional auto-reload tops your balance up at the moment it can no longer cover what is due, so a client signing up never hits an empty balance. The remainder of your most recent top-up is refundable.

How model costs work on a resold Hermes bot

Hosting and AI model usage are separate costs, and this is where reseller offers differ the most, so here is exactly how it works for Hermes today:

• Paid bots run on model funding from your own Build & Resell account, not on a key from your client. Price your package with model usage in mind.
• Free-trial bots are a Hermes-only option: the bot chats on a capped per-bot allowance funded by your own OpenCode Go key, and one API call moves it onto the paid route when the client converts. Our support team sets up the trial key with you before your first trial bot.
• Your client's own key is not available on Hermes yet. Handing a client's model key to a bot at create time is an OpenClaw-only feature, and the API refuses it for Hermes rather than quietly ignoring it.

So if your offer is “bring your own key, pay us only for the assistant”, run that line on OpenClaw (see white-label OpenClaw hosting). If your offer is an all-in monthly price, Hermes works well and the trial mode gives you a free tier that cannot run up an open-ended bill.

A simple pricing example

As an illustration only, say you sell a Hermes assistant for your niche at $29 a month. Your hosting cost for a Lite bot running all month is about $5. What is left has to cover model usage, your support time and your margin, and how much model usage a client generates depends on how hard they use the assistant. Start with a usage-capped package, watch a few real clients for a month, then set your tiers from what you measured. Your retail price is entirely yours; we never show it to anyone.

What you control, and what we run

One design choice worth knowing before you launch: deleting your storefront revokes its API keys and takes the shop down, but it does not delete the bots your clients are using. Those stay on your roster until you stop or delete each one on purpose, so an unrelated cleanup can never cut a paying client off by accident.

Before your first paying client

1. Create one bot, chat with it, delete it, and read the ledger. That round trip shows you the real economics.
2. Wire up the webhooks so your billing and support hear about state changes without polling.
3. Write your cancellation policy into your terms: stop and keep data, or delete and end the charge.
4. Turn on auto-reload, or keep a buffer, so a deploy never fails while a client is waiting.
5. Decide how model usage is priced into your packages before you quote anyone.

Frequently Asked Questions

Related Guides

• Build & Resell — the product page and a live storefront demo
• Build & Resell, in full — billing, limits and your clients' data
• Resell AI Agents — the API walkthrough step by step
• White-Label OpenClaw Hosting — the same model on OpenClaw, with client-owned keys
• Hermes Agent Team Hosting — one assistant shared by a whole team
• Managed Hermes Agent Hosting — hosting a Hermes bot for yourself