How to Install Hermes Agent and Start Your First Chat

Nous Research provides a desktop installer for Hermes Agent on Windows and Apple Silicon Macs, according to its installation guide. The same guide provides a separate installation path for Linux. This guide follows the official quickstart from installation to a working conversation, then shows how to connect Telegram for messages away from the computer.

Choose where the agent will run

Hermes Agent is the software that receives your request, uses an AI model and can carry out tasks through tools, as described in the Nous Research quickstart. You can first run it on a computer you already own. If you want to reach it while that computer is asleep, you need a machine that stays available. The Hermes messaging guide describes a gateway, the background service that keeps a chat connection open.

For Windows, the official installation guide offers the Hermes Desktop installer. The platform guide lists Windows 10 and 11 as supported systems. Nous Research also supports a direct Windows installation for people who prefer a terminal, according to its Windows guide.

On a Mac, the desktop installer is intended for Apple Silicon, according to the installation guide. Check the processor in the Mac's system information before downloading. For Linux, the same guide provides a terminal installer and lists the basic requirements. Follow the linked instructions on the day you install, since those requirements can change.

Both routes lead into the same first conversation flow.

Install and choose a model

Open the Hermes Desktop download page and choose the installer for your computer. The official installation guide says the complete desktop package includes the agent and the software it needs to run. Its first run setup then leads you through the model connection.

An AI model supplies the responses that Hermes uses. The Hermes quickstart says the setup asks you to choose a provider, then saves the choice before your first chat. A provider is the company or service that gives the agent access to a model. Follow the provider's sign in or key instructions shown during setup. Keep any key inside that setup screen and out of messages or screenshots.

If you installed the terminal version, the official quickstart sends you through its setup wizard before the first conversation. The command reference explains that its model setup screen can add a provider, complete sign in and choose the saved default. This is the point to check that the selected provider is one you intend to use, since model access may have its own charge or subscription terms.

You can keep the first setup simple. Choose one provider and one model, then test a conversation. The Nous Research quickstart recommends verifying a normal chat before adding a gateway, scheduled tasks or extra tools. That gives you a clear result to check at each step.

Send a first message you can verify

After setup, open the chat window and ask Hermes to describe a file or folder you have chosen. The official quickstart uses a question about the current folder as a first test. A useful first request is, “Tell me which files are in this folder and describe what each one appears to contain.” Choose a folder with material you are comfortable sharing with your selected model provider.

The quickstart says a successful first test shows the selected model and produces a reply. If you allowed a file reading tool, you can also check the reply against the files in that folder. This matters because a successful installation alone leaves the model connection untested.

If the chat does not reply, return to the official quickstart and check the saved provider, model and sign in. For the terminal installation, the Hermes command reference distinguishes the provider setup screen from the model switcher inside an existing chat. Use the setup screen when you need to add or reconnect a provider.

Once one ordinary chat works, try a small task with an obvious answer. Ask for a summary of a short document you selected, then compare the reply with the document. The quickstart recommends checking a real task before expanding the setup. A short test makes it easier to see whether the agent has the access it needs.

Add Telegram if you want to chat away from the computer

Hermes can connect to Telegram through its gateway, according to the Nous Research messaging guide. The gateway is a background process that receives a message and sends the response back to the chat app. The quickstart places gateway setup after the first working chat, so you can identify a messaging problem separately from a model problem.

Follow the official Telegram guide to create a Telegram bot and connect its credential to Hermes. Keep that credential private. The same guide explains how the gateway receives messages and how to check a personal chat before using groups.

For a bot that should answer while you are away, its gateway needs a running machine. The Telegram guide describes a default connection that keeps asking Telegram for new messages. A laptop that has gone to sleep cannot keep that connection active. This is why the location of the agent matters after the first desktop test.

If you want the agent on its own server, OneClickClaw offers a managed Hermes deployment on its Hermes page. The OneClickClaw setup guide says Hermes has its own dashboard and that the model is chosen from the AI Model tab on the server page. You can start with the seven day free trial and set up the model and chat connection on a server intended to stay online.

The order is simple: install Hermes, connect one model, check one real answer and then add the messaging channel you plan to use. Each step follows the Nous Research quickstart. At the end, you should be able to name the machine where Hermes runs and show one message it answered from your chosen chat app.