← Back

Hermes AI Agent Finally Gets a Desktop App So You Can Stop Using the Terminal

Original version · Jun 5, 1:30

The absolute darling of open-source AI is shedding its scary hacker skin. No more typing esoteric commands into a black box just to get some automated work done. This is the moment AI agents actually become usable for normal humans who don't live in a terminal.

The company Nous Research launched the official public preview of Hermes Desktop, bringing its powerhouse autonomous agent to macOS, Windows, and Linux. Previously, getting this 180,000-star GitHub sensation to run required messing with command lines, chat platforms, or shady third-party web wrappers. Now, it is a simple double-click install.

What makes this agent special is that it literally learns from its own successes by writing its own markdown skills and saving them into its long-term memory. It acts as a clever coordinator for major heavyweights like GPT-5.5 or DeepSeek-V4-Pro. Instead of wasting API tokens relearning how to format weekly reports, the app remembers the trick and builds a reusable skill for next time, meaning users do not have to explain the exact same task twice.

Under the hood, the new app uses Electron and React with a Python backend, meaning all project memories and custom setups sync perfectly from the old terminal core. While macOS and Windows users get nice, lazy installer files, Linux nerds are ironically still forced to use the terminal to set it up. The app connects to over 300 models, can delegate tasks to smaller sub-agents, and even sends you status updates directly via Telegram, Discord, or WhatsApp.

This move finally puts pressure on its main rival, OpenClaw, which had a cozy head start simply because it launched with a pretty face from day one.

It seems the era of bragging about running AI from a raw terminal is officially dead, as even the most hardcore open-source tools capitulate to the comforting warmth of a basic graphical user interface. True geeks will surely mourn the loss of their command-line superiority complex while secretly enjoying not having to debug launch scripts every Tuesday morning.

Source: Nous Research

Comments

This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.

3/24
  1. Frozen Raven
    finally, i can pretend im a prompt engineer without looking like a 90s cyber-goth
    +3 funnyFinally, a way to LARP as a tech genius without the aesthetic commitment of a trench coat and neon goggles