Hermes V0.14 is INSANE!

Hermes agent just dropped version 0.14 and this is the biggest free update they've ever shipped. Your AI agent is about to get dramatically faster, smarter and more powerful completely for free. We're talking browser control that's 180 times faster than before. Computer use now works with every major AI model and a one-line install that means anyone, and I mean anyone, can get Hermes running today in one click.

Plus here's the part that I really want to show you. There's a new way to stack Hermes with grok 4.3 that turns it into a full AI media studio. Images, videos, real-time Twitter search all inside one system all free. I'm going to show you the exact system I built with it step by step.

Plus there's one feature in this update that most people are completely sleeping on and it changes the way your agent handles memory. Once you see it you'll never go back to running AI the old way. So stick with me until the end because by the time we're done you're going to have the most powerful version of Hermes you've ever run and a full system around it that works for you 24 7. Let's get into it.

Hermes agent v 0.14 just dropped. This is a brand new free update to Hermes agent. You can see I'm on version 0.14 right here and this is a brand new update from Hermes. It's been released and I'm going to walk you through exactly what it means, how to use it, how to get it, etc.

Now if you want to know how to update your Hermes agent, all you do is you go into your terminal like this and you type in Hermes update and it's pretty straightforward. You can update from there, it'll pull in the latest updates and that's how you get the new release. You can also go inside Hermes chat and just ask it to install it from there right but the terminal command is by far the easiest and fastest way to do this. So what does this mean for you?

How does it work? etc. Let me walk you through everything that it means, what's dropped, what's new, etc. So if you go into Hermes agent, these are the new changes.

So the first one, the big feature is that you can install Hermes with one line, right? So before 0.14, you had to load the repository, run shell scripts, create dependencies, a lot of work for someone non-technical. Now you don't need to work when you use Hermes agent. All you do is you type here install Hermes agent terminal right now and then you've got it installed.

So it's super easy to set up now which is great and that means with one command you get the full experience, right? So you get the Hermes dashboard, the tools, everything that comes out of the box with Hermes and this means this is huge because it means that everyone can get started, right? You don't need to be a developer. You just need this terminal on the platform.

It's pretty simple and easy and bear in mind Hermes agent is the biggest open source agent in the world right now. It's overtaken open-source. So what this means essentially is that with this new update, even more people are going to use it, right? There's even less barriers to installing this easier than ever which is great.

Let's talk about the next feature and this is a big one for anyone using it which is grok 4.3 with a million token context window inside Hermes agent. So you can see for example I'm using Hermes with grok 4.3 here and you might be wondering okay what is a token? But you want to think of it like one word or one piece of word, right? So one million tokens means you can paste an entire code base, entire research document, or your whole business SAP library into one single prompt and Hermes can handle it.

And XAI, the company that makes grok, just connected to Hermes via aworf. So if you're already subscribed to Hermes, sorry to grok, then you can go inside the chat here. You can type in Hermes model like that and then you can actually change your model to use grok and you literally just it's an aworf. So that means you don't need an API, you just log in.

It's pretty simple to set up, right? And also it means if you're already subscribed to Twitter or to grok, you can use grok 4.3 inside Hermes without a separate API key and it unlocks some really powerful features. Let me show you an example of how this works, right? So if we go into this AgentOS system that we've built here, we can generate images, videos, speech, and we can even search Twitter for the latest updates, right?

So I've actually built out this AI studio using new updates from version 0.14, as you can see. So we can view our images here, we can use the videos that are plugged in, right? So we can generate videos with one single prompt and this is free to use, right? Because I've already subscribed to Twitter and I'm just adding it into Hermes agent and then we've got speech as well.

So we can generate speech with a command as well, which is pretty cool, right? Now if we want to search with the Twitter search here, you can see an example. So once I'd plugged in grok 4.3 into Hermes agent, you can see I can say search the latest AI automation news in x here and it'll pull in all the latest news from new updates inside Twitter, right? So it's a really powerful way to find like the latest news, new trending topics, it's great for content creation, social media posts, etc.

And we can pull in all of those updates right here. And then if we wanted to generate an image or a video, we can just say create an image of a dragon skateboarding, for example, and then it gives us the link to the generated image as you can see right here. Pretty cool stuff. And if we go inside the studio here, this is something custom that I built and this is an agent OS, right?

So basically I have a board, I have open core, I have Hermes all built in. You can get the link in the comments description if you want to get this inside the AI Puffer boardroom if you want the full system with all the prompts and the files I used to set this up. But basically this allows us to just use Hermes as a media content building system now, right? If I want to generate images, I've just given it a prompt, right?

If I want to generate videos, I can just give it a prompt. Let's take an example. So if I go into, I like to find out, actually we can do this inside Hermes. So I can say, give me a cool prompt or an amazing video about cats, right?

Just as a random example here. Now you could use this, for example, for Facebook ads or wherever you're posting social media posts, etc. And we can ask for the prompt here, it will come back to us with the prompt in a second. And then we can use that to generate a video as well.

We could also do the same inside inside Claude as well, but we've got the prompt here. So we can take that, then we can go down to our studio, go to videos, we want to generate this, boom, right? The reason that I would generate the video prompt with AI first is because you're going to get better outputs, like the AI understands how to talk to AI, right? Especially if it's already trained with the skills to generate these videos.

And so if I just type something in, it's going to take a lot of time, going to take a lot of my mental energy to come up with it. Whereas I can just ask Hermes, and it already understands grok, because grok is already plugged into it. And then we can generate the best possible outputs. And that will start generating the background here, you can see it's generating now.

Let's see what we got from the questions here. So DJ says, Good evening, good to see you here. Matt says, Thanks, happy to help. Not sure on that one on I wouldn't recommend it.

Basically, that's not something that I focus on. But welcome says, I'm using runway for video generation. What I want to know is how to how do you make those long videos. So we have a full section inside our community here on how to do that, right.

So if you go to the classroom, and then you go over here, you can learn how I create my videos with this full process, we've got a full step by step operating procedure that you or your team can use, and a full video training right there. So if we go back to agent OS, now, we can see that's generating video, we can come back in a second. Let's talk about what else is new in the meantime, in terms of what's new inside Hermes with this update. So the next big feature is the local open AI compatible proxy, right.

And this one of the most powerful features in this update. So you can run one command. And this creates like a local server on your computer. That's not server speaks the opening AI API language.

So any tool that works with opening AI, right, for example, like client could be codec, CLI, etc, you can use your existing subscription to operate that, right, there's no API key, which is pretty cool, right, one subscription, you get every tool. That means for example, let's say for example, you have a chat GPT subscription, you can get it to use across every single coding tool that you already have, which is pretty powerful. Then we have the Twitter search built in. So like I mentioned, the agent can now search x directly, no skill install required is already built in to Hermes agent, you don't need any custom integrations, you don't need any API's, which is great as well.

So you can just ask your agent search x, and it does it right, it can find threads, it can surface posts, it can track trending topics straight from the chat as well, which is pretty awesome. Now you can see that video has generated inside the chat that we talked about a second ago, it does have audio over the top. And also, when it's playing here, you can see that we can preview it inside our studio, which is pretty cool, right. And so Hermes agent came up with the prompt, we plug that into the studio, we can generate videos now we've grown 4.3 as well, which is pretty amazing.

We can go full screen on that as well. And that's stored locally, which is pretty awesome. We can also save that. So we can download that as an MP4, as you can see here, it's pretty simple and easy to create those videos with AI, pretty powerful stuff.

And then also what I like about this is like, if you're just using the terminal, and you generate a video, you're not gonna have any histories, you can't really look through the chat, you can't really go back and see visually what it looks like, right, you actually have to open up separately. So I'd recommend that you have an agent OS system like this, so that you can see the videos, you can preview them inside a nice studio before you and if you need to track anything, you can, you've got your image history, video history, everything, right, it's pretty awesome. Next up, we have the Microsoft Team Support. So your agent can now read and post in Microsoft Teams with a full Microsoft Graph stack.

So off web books, event listening, outbound messages, they're all wired up inside Hermes. So if you run your team, and your team runs on Teams, your agent is now in the room too, if you want to add it, which is pretty cool. There's another big feature here, which is 180x faster browser control, which is absolutely crazy. So when your AI agents were browsing the web before, every single action needed a new browser connection, that's pretty slow, right.

Now Hermes routes everything through one persistent Chrome connection, and so stuff that used to take like over two seconds can now return in milliseconds, and you can use this pretty fast and easy, right, which is pretty awesome. So if we wanted to use this as an example, I can go inside Hermes here, and I can say, okay, go using or use Chrome, and go to Google for me, right, as an example right there, and then it will just start loading, and we can use the browser control right there, which is pretty awesome. I'll show you that in a second, because it would just take a couple of minutes to load the skill in, right. You can see it's actually browsing right now using this, right, so you can easily like navigate, and navigate right there, which is pretty awesome.

Next up, we have faster startup time. So every time you launched Hermes before, it loaded everything up front, right. Now it only loads what you actually use, so startup went from slow to near instant, right, and the tool screen load went from 14 seconds to 1.5 seconds. So if we load up Hermes here inside the chat, you can see it just loads super fast, which is great.

I don't know if you ever used something like OpenClaw, for example, other agents like that, but they can be quite slow to use, whereas with this new update, it's just a lot faster to load everything, which is great. You've also got computer use on any vision model, right, so previously the computer use feature, where the agent controls your mouse and keyboard, only worked with Anthropx models, right, now works with any vision capable model. So this could be like GPT-5, could be Grub, could be Gemini, could be Claude, could be all of them, right. Your agent can now see your screen, and operate your computer, and open up apps, regardless of which model you're using.

Let me show you an example of this. So if we go inside the chat here, and we say, okay, use computer use to open up and write a new note inside my notes app, and what it will actually do here, is it will start loading up computer use agents, and just controlling my apps on my desktop. The other cool thing about computer use agents with Hermes v0.14, is that it's something that can work in the background, right, so it doesn't get, it doesn't distract me, it doesn't stop me what I'm doing, but it will just use my computer in the background, whilst I'm talking to you, right, whilst I'm doing something else on screen, and the cool thing about that is, like, it doesn't disrupt you, what you're doing, or how you're working, etc, and it's a lot faster, and also, like, if it's a slower task, no problem, because it's just going to work in the background, and it's not going to distract you, or slow you down from doing anything else, whilst that's working. As you can see here, it's using computers, using notes, and it's going to start creating a new note in the background, so we can see computer use running right there.

What else we got? We got the live session handoff, so let's say, for example, you're mid-conversation with a fast model, and you want to switch to a deeper reasoning model, without losing context, before you lost everything, and had to start over. Now, you can type forward slash handoff, sorry, forward slash handoff, and that moves your entire session, every message, every tool call, every piece of context, to the new model live, right, so you don't need to restart the gateway, you don't lose any context, or anything like that. By the way, you can see here, the Hermes has just finished using computer use, so it says notes app opened, using computer use, new note created in Apple Notes, if it goes to Apple Notes here, it's created a new note.

We can say, open that up for me, and again, it will use computer use to do that, and then you also got native Windows beta, right, so Hermes now runs on Windows without WSL, so you don't need Linux, or anything like that, you can just install, and go from there, which is pretty cool. In the, as well, there's nine new big skills that I'll come on to in a second, so these are add-on modules you can install for specific jobs, which is pretty interesting, so you got some interesting stuff that you can do right here, your API testing, an evolutionary prompt tuning system called Darwinian Evolver, which is pretty great, that in itself could be a whole different tutorial right there, I think. Shreya says, love you man, admire your drive, thank you very much sir, I really appreciate that. Next up, we have Discord channel history backfill, so when your agent joins a Discord channel for the first time, it can now read recent message history, which means it just has more context, right, it knows what's been said, so what this means, essentially, is if you've been talking about something, but you don't want to re-explain that whole context to your AI agent, it can just go off, and take the latest messages from your channel, which is super useful as well.

We have a native button UI on Telegram and Discord, so when an AI agent asks you multiple choice questions, it shows real buttons, you can tap the button, the agent gets the answer, you don't need to type anything like that, which is pretty cool. There's some other stuff as well, like for example, I think this is not applicable to 99% of people, but you can see there's some semantic diagnostics improved on every file right, and also per turn file verifier as well, but I think for 99% of people, it's not that useful, right. Now you can see, for example, this is the note that Hermes agent wrote, so we said, for example, create a new note about call automation, so I could implement today, and just whilst I was talking to you in the background, it created this new note, as you can see, the Hermes has come up with, and the cool thing about this is, basically what it's done here, is said, okay, based on your current setup, and it has all this information about me, it can basically come up with ideas personalized to me, right, and it takes that all from the memory system that I built in, which I'll come on to in a second. So really powerful update, lots of cool new features, I really like the computer use, the browser use agents, the AI studio that we have as well, with grok 4.3 implemented, more powerful than ever, easier to use than ever, and it's just a really cool update.

Now you might be wondering, okay, this stuff seems techie, AI tools are just for tech people only, or how confusing or difficult is it to use Hermes agent, but with the new updates, you can just install it with one command, right, you just do pip install Hermes agent, that's it, so you don't need to do anything technical, if you can type one line, you can get started with Hermes agent today. Other people say, I don't know which AI model to use, and so I wouldn't worry about that, like the stack that I'm showing you today, lets you use every major AI model in one place, right, for example, if we have a look at, for example, the agentic OS, if I'm like, okay, today I want to use Hermes, I can just use Hermes inside the chat here, if you're like, okay, I want to use Claude, I really like Claude, no problem, you can switch to Claude over here, and inside the agent OS, right, it's way better than, for example, having to click over to Claude here, and then having to click into terminal over here with Hermes agent, right, that's a lot more messy, whereas this is like way cleaner and easier to do. Other people say, AI agents are too slow to be practical, but honestly, with this new update, like the startup time has been cut by 19 seconds, and browser actions are like 180 times faster, right, these are not small improvements, this is really the difference between a tool that feels clunky, while it feels like an extension of your brain, essentially, it's got that full context on you and everything else. Now, I want to talk you through the stack that I'm using right now, and the ideal setup that I think works for everyone, right, and this is the Goldie agentic command stack, right, which is a new paradigm for running agents inside your business.

Most people think about AI the wrong way, they think I'll use chat GPT for this, Claude for that, maybe Gemini over here, etc, but it's like hiring 10 different assistants who never talk to each other. The Goldie agent command stack that I'm showing you today flips that model entirely, and it's really powerful now that the new updates from Hermes have just come out. Let me show you an example, right, so you can now plug in whatever models you want, you got OpenCore, you got Hermes, any new updates or new agents that come in, you can put inside this agentic OS, and any sort of tools that you do day-to-day, you can also use often as well. So, for example, we have a goals section, we have the SEO agent section, we have the media studio setup over here as well, we have the journals, and we've also got a memory, which I'll come on to in a second, and we've got a full build guide in case, like, I need to know how to come back to that, right, and so I can manage everything with one command center using this agent OS system, and so you've got the brain, and this is the model that's doing the thinking.

So, in Hermes agent, you could plug in, for example, like grok 4.3, and you can pick the brain based on the job, right, so if it's like a big deep thinking job, you might actually switch over to Claude, right, but if you need something that's fast, then you could use something like, for example, a step 3.5 flash inside Hermes agent, which is free to use, right, then you've got the messaging platforms as well that you can plug in, which is layer 2, so you can have your agent inside telegram, or discord, or slack, or whatsapp, and just chat to it on your phone like a human, and it responds like a team of experts. You also got the hands, right, and these are the tools and the skills where Hermes actually does the actual work. So, with Hermes version 0.14, you've got web search, you've got twitter search, you've got browser control, which is faster than ever, you've got computer use, which created that note for us a second ago, you got the video generation section inside the studio here, you have the image generation, you could also connect, for example, any apps that you want, and you've got all the new skills I talked about as well. Now, you also have the memory section, and this is probably the part that most people miss entirely, but it makes your agent a hundred times smarter, right, literally way more powerful.

The reason that we can get such good automation ideas personalized to me right here is because we've got this memory system built in. So, what is a memory system? This is basically linked to my obsidian vault, and so if we go inside here, I call this the infinite context engine because it has information about me, the tools I use, what I do day-to-day, my typical tasks, any sort of people in my life, or working for my agency, etc. We have, for example, all the information on like, even like the airlines I fly in and stuff like that, or like recent pieces of content, websites I own, etc.

And everything's linked together, so the AI agents that are connected to it, for example, like Hermes, understand me, my context, and everything else. So, if I go into Hermes and I say, okay, give me some ideas on automations that I could build today, it'll give me recommendations, as you can see right here, right, and so it's very personalized to me, and it's way more powerful. If your agent doesn't understand you, or your context, or who you are, etc., it's not going to be useful because the problem is it doesn't understand what you do all day, right, and if it doesn't understand what you do all day, how is it going to understand how to help you, right? That's the biggest part here.

So, what we do is we link Obsidian as a memory system, and my agents take notes for me all day, and then that's plugged into Obsidian, and my agents can just take the local files from Obsidian directly. Here's an example, so you can see here that we have notes on everything that I do all day, everything inside my team. You've got, for example, even notes on how I typically structure funnels and stuff like that. It's a really powerful system that all links together, is all automatically color-coded, and super powerful to context, and then Hermes has that information, OpenCore has that information, and any other AI agent that comes out in the future has that context, too, because we can just plug it in, right, and so your agent remembers context from the last session, your instructions carry over, your responses come back faster, and also, bear in mind, like, you're going to be using less tokens to use this stuff because if the agent requires less prompting to get more personalized outputs, then you don't have to prompt it as much, you don't have to go back and forth with it, so it saves time with you, it also saves tokens on, like, the amount you would use it, right, because you get powerful prompts back first time around instead of later on.

We have a full system inside the AI Profitable Environment, it's called the Infinite Context Engine, and that's exactly how we set this up, if you want to learn more about that, and then layer five is the command center, so this is the mission control, which you can see here, and this is where you manage everything, so in the Goldie Agency command stack, this is your dashboard, essentially, you see all your agents, you see all your running tasks, you can manage your multi-agent Kanban boards inside there, you can manage all the skills that we've set up, we can see what activities we worked on recently as well, which is super useful, and you can run scheduled tasks as well, which is great, right, so everything is inside there, everything that we need is ready to go, and we can just use that anytime we need to, which is great, and when you, the thing that I would say here is, like, when you stack all five layers together, you don't just have an AI tool, you have an AI operating system that's running everything for you, and this is the command stack, right, and Hermes Agent v0.14 made every single layer more powerful, so that's basically it for the updates, we've covered what might be holding you back, and how to fix that, we've covered how to set up the update, we've covered what the update looks like, and how you can use it, and we've also covered a full guide, right, so just to recap here, quick recap, what is Hermes Agent? It's an open source AI agent platform that connects your AI models to messaging platforms, tools, skills, and automations, so you can basically automate anything with it. What's new in v0.14? So you've got the one-line install, grok 4.3, 1 million token context window, local OpenAI compatible proxies, Twitter search, Microsoft team support, 180x faster browser control, way faster startup, computer use on any vision model, live session handoffs, native Windows beta, nine new skill packs, and that's basically it.

What is the Goldie-Agentic command stack? So that's basically, you've got the brain, which is your model, your platforms, the hands, which your tools and skills, the memory, and then also the command center, which you can see right here. So if you want the full guide and everything, I'm going to put that inside the AI Profit Boardroom over here, and I add new tutorials inside here every single day, so we're going to add a new tutorial on all the updates inside v0.14, and also how to plug that into the AgentOS system, and there will also be a video tutorial on that stuff as well, right, and you can see we update this every single day with new tutorials, new updates, we've got the full Hermes AgentOS system here as well, it's just an amazing community of 3,000 business owners learning, growing, helping each other, and also inside the calendar you can jump a weekly coaching course, you can get all of my best new trainings inside the classroom here, you can connect with people in your local city from the map, and that's all inside here, right, link in the comments description, or go to the aiprofitboardroom.com to get access. Let's see, Michelle says, did you successfully get Paperclip connected to Hermes?

Yeah, I did, I did get it set up, honestly I don't use it that much, I think the 99% of people, unless you need a huge team of agents, you don't need Paperclip set up, I think it overcomplicates stuff for most people, and also a lot of people have trouble like integrating it. I much prefer to have a setup like this with the AgenticOS, where I can just switch between my agents and they can work together, and then any stuff that I do day-to-day, for example like managing my goals, SEO, or the studio, or the journal, etc, it's all inside there ready for me to go, right, and I think it's a much nicer UI as well. Chad says, thank you, happy to help, how is the stability of the dashboard? Really good, so you can see here that it works pretty quickly, it always loads, it's always ready to go, it's hosted locally, so that's pretty useful as well, and then if we want to, for example, switch between different tasks, what's happening, etc, we can do that pretty simply here, right, like it works pretty well, it loads pretty nicely, it's quick and easy to use, etc, so if I'm like, okay, I want to add a goal in here, I can just type that in, so I can say, okay, setup Hermes v0.14, right, we add that goal, we can tick it off, it works nicely, right, it's really responsive, and also it stores your history here, so for example, like these are recently deployed blogs that were built, we can see the history here, we can go and check the blogs that we've created, quality control them, etc, pretty easy and simple to navigate around.

Which agent do you prefer, OpenCore or Hermes? I much prefer Hermes, I just find it smoother, easier to use, I also find, like, when they release updates, it works really smoothly, right, so when I set up, for example, Grok 4.3 with the new video studio, the images, etc, it just works perfectly, straight out of the box, it is really nice and easy to use. Chaz says, I'm just using Hermes right now, doing web dev, does it make sense for me to install it? It depends, like, what I would do is, inside the setup here, I would add a new layer that's called web dev on the side, inside the agent os, and I would just say, okay, add a new agent on the left hand side that says web dev, and then you can come back and manage all your projects in one place, so for that reason, yeah, I would use it.