NEW Hermes Web UI Looks Too Powerful

Ajouté : 2026-05-11

[00:00:00]New Hermes web UI looks too powerful.

[00:00:03]Here's everything it can do. What if your AI agent never had to stop working just because you closed your laptop?

[00:00:09]What if it remembered everything, learned from every task, and got smarter the longer it ran? What if you could chat with it from your browser, your phone, your terminal, all at the same time? Most people are still using AI like a calculator. Type something in, get something out, done. But there's a whole different level. And today we're going there. Hey, I'm Amelia, the digital avatar here to help you actually learn and use AI tools in your real work. Not just talk about them, use them. Today, we're covering Hermes web UI, a browser interface for the Hermes agent built by Noose Research. I want to show you exactly what it does, how it works, and why it's worth your attention right now. Stay with me. Let's go. So, first, what is Hermes agent? Hermes Agent is an open-source autonomous AI agent built by Noose Research, the lab behind the Hermes model family, Nomos and Psyche. The tagline is the agent that grows with you. And that's not marketing. This is not a chatbot. This is not a coding assistant you only use inside an editor. Hermes is designed to live on your server, keep running, learn your projects over time, and build its own skills automatically. It has persistent memory. It doesn't forget what you worked on last week. It supports scheduled automations, tasks that run on a crone schedule in plain language, unattended through its gateway. It delegates to isolated sub aents, smaller agents running in parallel, each with their own workspace and context. It supports five sandboxing backends, local, Docker, SSH, Singularity, and Modal. And it works across 17 different messaging platforms.

[00:01:54]Telegram, Discord, Slack, WhatsApp, Signal, Email, CLI, and more. Start a conversation on one, continue it on another. This is a real agent, not a wrapper, not a demo. Now, let's talk about the web UI. No build step, no framework, no bundler, just Python and vanilla JavaScript. Straight from the project readme. Fast to set up, nothing complicated. The layout is three panel.

[00:02:21]The Redmi actually describes it as a clawed style layout. Left sidebar for sessions and tools, center for chat, right for your workspace files, and it has full parody with the terminal experience. Everything you can do from the command line, you can do from this UI. Here are the actual features.

[00:02:39]Responses stream in real time using server sent events. You can connect to multiple AI providers. Open AAI, Anthropic, Google, Deepseek, Newsportal, Open Router, and the model dropdown populates dynamically from your configured API keys. Switching models is one click. You can send a message while the agent is still working on a previous one. It cues automatically. Edit any past message in line and regenerate from that point. Retry the last response with one click. Cancel a running task from the activity bar. Tool call cards show up in line. the tool name, the arguments it passed, a snippet of the result. You can see exactly what the agent is doing at every step. Mermaid diagrams render in line too. Flowcharts, sequence diagrams, Gant charts. If the agent generates one, it shows up rendered directly in the chat. There's an approval card for dangerous shell commands. The agent stops and asks you allow once, allow for the session, always allow, or deny. you stay in control and the connection autoreonnects on network blips which the documentation flags specifically for SSH tunnel resilience. There are six panels in the sidebar total. The tasks panel is where you manage cron jobs. View, create, edit, run, pause, resume, and delete scheduled automations. When a job finishes, you get a toast notification and an unread badge on the tasks tab, even if you're in a different session.

[00:04:10]The skills panel shows all the agents skills by category. Search, preview, create, edit, or delete them. Skills are how the agent stores and reuses learned capabilities. The memory panel lets you view and edit memory MD and user.m MD directly in the browser. No need to SSH in and edit files by hand. The todos panel shows a live task list for the current session. The spaces panel lets you manage and switch between workspaces from the top bar. When I first started diving into agents like this, I was honestly overwhelmed. The tools, the configurations, the different ways to set things up. That's when I created a community called AI Profit Boardroom.

[00:04:55]Over 2,000 members all focused on learning AI together and sharing what actually works. It taught me which workflows save time versus which ones waste it. The community shares real use cases and practical implementations. If you're serious about using AI to improve your work and skills, check it out. Link in description. Getting it running is straightforward. First, install Hermes agent. One curl command from Hermes agentnews.com link in the description. Once Hermes is configured, the web UI is three commands. Clone the repository. Navigate into the folder. Run start.sh. SH the script finds your Hermes installation automatically, sets up the Python environment, starts the server, and prints the URL on a remote machine. It also prints the exact SSH tunnel command to access it securely from your browser.

[00:05:48]Default port is 8787.

[00:05:51]There's also a native Mac OS app. The Hermes Swift Mac app is built with Swift. No Electron, no heavy dependencies. It wraps the web UI in a native Mac OS window with a doc icon and menu bar created by a contributor called Red Spark Labs, now maintained under the Hermes WebY GitHub organization. It supports direct mode for local instances and SSH tunnel mode for remote servers, managing the tunnel automatically on launch and tearown. You get clipboard integration for pasting text and images with commandV, file upload via a paperclip button, Mac OS notifications when responses finish in the background, voice input, and auto updates via Sparkle. Signed and notorized by Apple.

[00:06:36]No gatekeeper warning. Latest version is V1.0.9 April 2026. Download the DMG, drag to applications, and it connects to localhost 8787 by default. For remote servers, open preferences with command, comma, set ssh tunnel mode, fill in your details, and it handles the rest. Hermes also has an official web dashboard. Run Hermes dashboard in the terminal. It shows your installation overview, recent sessions with model info and message counts, and embeds the full TUI in the browser. Runs entirely on your machine.

[00:07:12]Bottom line, Hermes agent lives on your server, builds persistent memory, runs on a schedule, delegates to sub agents, and reaches you across 17 platforms. The web UI puts all of that in a clean browser interface with no build step and full terminal par. The native Mac app wraps it in a proper desktop experience, and the whole thing is open- source under the MIT license. The gap between chatting with AI and having an agent actually do work for you is closing fast. Hermes is one of the clearest examples of that right now. If you're looking to dive deeper into AI tools and actually implement them in your work, I recommend AI Profit Boardroom. Over 2,000 people learning how to use AI effectively. Everyone shares real experiences. What's working? What's not?

[00:08:00]Which tools are worth your time? Which ones to skip? No hype. just solid practical guidance from people doing the work. Link in description. If you want the full process, SOPs, and 100 plus AI use cases like this one, join the AI success lab. Links in the comments and description. You'll get all the video notes from there, plus access to our community of 58,000 members who are crushing it with AI.