{
  "video_id": "L0tQrTQBmjI",
  "segment_count": 1291,
  "duration": "41:22",
  "full_text": "Hermes is the world's most powerful AI agent, but if you're new, it can be incredibly overwhelming. You're probably looking at the app and asking, \"How does this make my life easier? What is a habit? What is an operating system? How does this work?\" And without knowing the basics, you will never unlock its full capabilities. So, in this video, I'm going to give you the exact tutorial I wish I had when I first started. We'll break down 21 Hermes concepts, starting from extremely simple building to its most powerful features. And by the end of the video, you'll understand this incredibly powerful technology, even if you're a complete beginner. And if you're new, I'm Jack. I built and sold my last tech startup with a gazillion customers. Now, I'm building my own AI startups, and I share here the stuff that actually works. So, if you haven't already, grab that beautiful coffee, and let's dive straight in. So, let's begin. Every Hermes concept explained for normal people. Now, this is going to be the Hermes tutorial. If you just watch this video, you will be able to walk into any room, and people will think, \"Dude, who the hell is this guy? Who is this Hermes genius? Does he work for Hermes?\" That's what you They're going to feel at the end of the video. Now, Hermes is massive. 271 billion tokens have used, which is around about 240 billion, 230 billion words. It's across 22 messaging platforms. It is huge. First concept is the foundation, which is basically it is what we call an agent, not a chatbot. This is fundamental to how you understand and think about the Hermes agent. A chatbot tells you how to book a flight. An agent will actually go ahead and book the flight and find it for you. So, think about this. With a chatbot like ChatGPT and Claude, you give it a goal, and it creates a plan. But with an agent, it has tools, and it can actually go ahead and do actions for you. So, for example, what a lot of people don't really understand here, if you think about it like this, right? It's like a smart friend versus is personal assistant. Hermes is a personal assistant. Let me show you what I mean. So, here for example, I have the Hermes agent connected. If you haven't set it up and you want to learn that, I'll put a link on screen to check out. I can say, \"Hey there, my man. I'd like to find for me the cheapest flight from Dubai to Toronto. Flying out, let's say, at any point within the next 2 weeks. Let me know what that looks like.\" You can ask it questions and this will go ahead and use its tools, its agentic capabilities to take actions for me, find information, and enables me to do anything that I like with it. And so, it's done a manual check. I then said, \"Find me one flight and open it up for me in the HTML.\" It looks gorgeous. It went ahead and did that and as you can see, I now have it there. So, conceptually, Hermes is an AI with access to tools. Could be your Gmail, your calendar, many different things that makes it so powerful. So, wherever you are, wherever your phone lives, you can enact things, you can affect change with your AI equipped with these tools. The second concept of Hermes is what it actually is and when you would use it versus using Cloud Code, ChatGPT, Codex, Grok, all these different things. I want you to think of them in this as a framework to understand where Hermes fits in with this. So, think about this. Hermes is like your beautiful Labrador. It's your dog. Cloud Code is your contractor. Um OpenFlow itself is your roommate. And then, Antigravity can be like your your IDE. It's your it's your buddy, the guy that you work with. The best way to think about this is if Hermes is your dog, it lives with you. It knows you. It gets smarter year over year. And Cloud Code is a contractor. It's brilliant for a specific job, but it doesn't necessarily remember you. The core thing about Hermes and the whole point of it is that the more that you use it, the more it gets to know you, the better it gets, and the more specific it is with you. And what it basically means is you can use it wherever you go, wherever your your mobile, you're in a coffee shop, a gym, you can whip out your phone, chat to an agent whose memory is always always improving about you. Now, can I have, for example, Cloud Code learn stuff about me, too? Of course, you can. We can get those memory systems. But I did with Hermes is we can take it anywhere where there's an it enables us to be fully mobile, fully custom, and just unlock some incredible capabilities. So, if I'm at the desktop, I might be using Claude, but if I'm out and about, I'm going to be using Hermes because the more you use it, the better it understands you. And its whole design ethos it's built to live with you and improve with you over time. Which brings us on to, importantly, the third concept, which is one brand, 22 mouths. What do I mean about that? Hermes itself can be accessed from over 22 different interfaces. We use Telegram, but for example, we could talk to it on Discord. We could talk to it on WhatsApp. We could talk to it on any different interface. You can even talk to it on different operating systems. For example, this is my Hermes operating system that connects to my Claude code operating system. And I can ask it questions. I'm like, \"Hey there, man. What day is it today?\" And I'm just speaking directly to my Hermes agent. I can see every single chat I've had on the left-hand side. And with Hermes, you can actually access the same brand, the same intelligence from all of these different locations. You aren't just constrained to one single input because where the Hermes is running, whether on your computer or VPS, it's the same intelligence and we can connect to it all these different interfaces. For example, if you come and use the Hermes setup code, for example, you'll be able to see all of the different locations that you can actually use with Hermes. For example, this shows you all the different message platforms Slack, Matrix, and all you can do is scroll through these, select them, and then you can connect and chat to the same AI intelligence from any platform that you like. You aren't locked in to just one software. And even in this one here, for example, if you download this Claude code operating system Hermes operating system, you can chat to it in your browser, which I think is a bit more of a dynamic interface. And so, if you think about it as as an analogy, that imagine that you have an operator sitting in the middle, and you can call into this operator from any different one of these interfaces, and she will route every single one of them to you guessed it, your Hermes agent. Which takes us critically onto the next one, which is going to be constant number four, which is where Hermes lives. Now, you've probably heard of it a few different terms like the terminal and VPS and locally hosted. And this is what's really important to understand. The first one is your own computer, which you can run completely for free. This is the easiest, it is the quickest, and in some respects, it's actually incredibly safe to do it this way rather than on other options that I'll show you in a second. You don't need to go and buy some crazy host somewhere. You can do it completely for free on your computer. To do that, all you literally need to do is come over to the Hermes website or click the link below. You copy this, you open up the terminal, and the terminal is basically the way you talk to your computer. It accepts code. And to pull it up, you do command and space bar, you type in terminal, and the terminal will appear. And all you do is control V that command, and effectively, that will install Hermes into your computer. Alternatively, you can go to an app like Cloud Code, and you can say, \"Hey there, I'd like you to install Hermes, please. This is the install code. Go and grab it from the GitHub.\" Or whatever you want to, right? And you can paste it in like that. And literally, Cloud Code will install it for you. So, this is what we mean when we talk about the terminal. Now, it running on your computer just means that as long as your computer is on, Hermes will run. And this is why lots of people have got either like, you know, old MacBooks that they run it on or or MacBook desktop or different kind of laptop, and it just runs for you 24/7 on that computer, or it can just exist on your computer. But effectively, if the laptop shuts down, Hermes doesn't work. So, that's effectively this local hosting revolution. I'm a big believer in local hosting. It's here. It's in my room. It's It's here. I can I can touch it. I can feel it. The second one is a VPS. So, this is essentially where instead of running on your computer, it's actually running on someone else's computer somewhere else in the world. And these companies called hosting providers effectively say, \"Hey, what we're going to do is buy like thousands of these like, you know, computers like this, and we're going to stack them up or whatever they're doing over there, and you're basically renting a virtual private server.\" Okay, so it's running over there. But one of the risks with this is the fact that it it's it's running somewhere else. They can be attacked. You need to do certain things to certify and protect the ports, and you pay monthly for renting that service. For example, our speech attacks at Glyder, that lets me say things and speak as if for instance, you know, we ran some space and for inference to make that fast. So, VPS is an interesting one, and you'll pay monthly for that. And that's what they mean when they say virtual private server. It's running remotely somewhere else. And so, just be aware that these virtual private server companies do pay people money if you sign up using links. So, it's not always necessarily the best thing. Now, this is run here on your computer and free. So, for example, if I come over here, I can say, \"Hey there, I'd like to shut down my Hermes agent, please.\" And you can literally chat to Claude to basically do different things. Now, you'll see as it does that, you'll see my Hermes button now will come down and say, \"Gateway is shutting down\" as this process goes down and closes for us. And as you can see, just like that, Hermes says, \"Gateway shutting down. Your current task will be interrupted.\" And it shut down, and I can say, \"Awesome. Go ahead and reboot him up for me, please.\" And you can do the same thing. And this can happen because it's running on my own actual computer, which is on 24/7. And this leads us on to the next and fifth concept, which is OAuth versus API. What the hell does that mean? So, there's two ways that you can connect to something essentially. One is what we call open authentication. It is dead straight or forward, and all it basically does is opens up a new browser, and you sign in. Let me show you exactly what I mean. So, here we go. We've got, for example, a terminal. So, we can chat to Hermes. Might come down on my dashboard on the left-hand side to have a look at the Hermes setup command. Let me just go ahead and copy this, throw this in the terminal, and you'll see what I'm in. So, as you can see, it's actually popped up and said Grok, Super Grok, and Premium. Now, for example, if you look at models, which come to shortly, or any servers, we could either do an OAuth. So, for example, if I come down here and I press enter, I might come down and reauthenticate. So, let me come down here and press number two, and you'll see what I'm in. I just want to show you, rather than explaining that. Look, this has opened up a new window. This is open authentication. So, instead of me giving it keys, it just says, \"Cool, well, you're signed in, right, bro?\" So, why don't you just go ahead and allow me, so you can literally come over here, you click on allow, and then this will now be fully connected to Hermes. Connection successful, and it is easy as that. Now, we're fully connected, and it lets you pick the model. How cool is that, right? Now, if we come and do that again, little hack, by the way, if you press up on terminal, it will actually show you the last thing that you messaged it, which is Hermes setup. And let's say we go now and do a different one. Let's say we want to do OpenRouter. I press spacebar to select it. Okay, come down. It's going to say active provider, and it's asking here for an OpenRouter API key. So, what this means is it needs a physical key. So, you come over to openrouter.ai/models, and from here, we're going to grab a key, and then input it basically here. So, those are the two ways you're ever going to connect. One is open authentication, and the second one is API key. Now, OAuth itself, you can use basically Grok, which is incredible, and you can use ChatGPT, but you cannot use Claude. With OAuth and API key, you can access all the different models, which I'll touch on very shortly. Best way to think about it is that an OAuth is like a sign-in. Okay, so one is a button that you can take back, and the other is a string of characters that live on a server somewhere, and anyone that use an API key with, you can just rotate it anytime you want to. And then once you rotate it in that environment, which is what OpenRouter is so good, it will never work again. So, that's like the big difference. And this takes it on very nicely onto phase two, and actually concept six, which is choosing the right model that that powers us. So, if you think of Hermes as a series of systems, okay? And I'll show you this down and show you exactly what I mean. Hermes itself, right, is an agentic framework with tools and capabilities. But, powering Hermes is a brain. So, for example, in Hermes, I can do {forward slash} type in model like so, and it will let me change the model, the brain behind Hermes. So, at the moment, it's using Grok, but I could switch to Open Router, which gives me access to all of the models on the planet. I can get to Anthropic, I can get to OpenAI Codex, I can do any of these things that I want to, GPT 5.5. Now, if you're using Hermes, you must connect your ChatGPT account to it. And the reason for that is you can use your $20 subscription to code. You cannot do that with Claude. You have to pay for API credits with Claude. So, you're going to see that GPT 5.5 is like the one of the best ways you can use it there a lot using your existing subscription, and it's exactly the same with Grok, which is why it's so freaking cool. Now, the idea here is that we want to pick the right tool for the right job. I am model agnostic. I believe in the multi-brain strategy, which is basically model agnosticism, or we're agnostic. And effectively, we use the best model for the specific task. So, the best model at reasoning right now, we believe, is Opus 4.7. So, what I do when I want to reason is I come over, I change the model like so, okay? And I will just go ahead and chat to this on a daily basis. I click on Open Router, and I pick Opus 4.7. Now, I pay for credits on this, but in Open Router, when you create an API key, you can set I want this to be $10 maximum per month. That's it. And if it hits $10, it just stops working, and you can go ahead and switch the models, okay? Then you've got, for example, other models that can do volume. So, GPT is an amazing generalist. Grok, again, you can use for volume. Grok, when you authenticate with your X account, can also search Twitter, as well. And then you can bring in other models like Deep Seek, and actually run your entire thing completely for free using a series of free models if that's what you wanted to. So, I want you to think about your brain more as a toolbox rather than a one-size-fits-all because to a hammer everything is a nail, but sometimes you want to use different things like Claude for design and ChatGPT for reviewing code and being more general activities. So, we're just going to tag in the models that we want to and to do that we typically want to use something like OpenRouter because that's one connection that gives you hundreds of different models. little fact by the way, if you come over to openrouter.ai/rankings, you can see month on month by the way who where all where all the things are being used. Look at this. Deepseek V4 Flash, right? That's roughly The Flash model is not as powerful as the main model, but effectively you can get like 95% performance for 1% of the cost. That's kind of how cool these different systems are. And Concept Seven is a data of what we call local hosted. And local hosted just means that not only is Hermes running your laptop, but the actual model itself is running on your own computer. Now, the trade-off for this is that the biggest models are billions and billions of parameters, which means you need data centers that are huge to run them. So, you're going to be limited based on that, but what you can do on your MacBook, for example, is click the Apple icon at the top, click on about this Mac, and you can see some information about it. So, you can basically screenshot that Mac, you can push that image into Claude or Hermes and say, \"Hey, for example, if I come down here and do this real quick, I screenshot this, I share it to Hermes, I can paste the image in and be like, \"Hey, based on my MacBook specifications, what do you think would be the most powerful local model that I could run?\" Okay? And what we can do that is actually download that model using something called Ollama that lets you download stuff. And you can even do Ollama Cloud, which is a free model hosted somewhere, or I can download it specifically like Gemma 4, Qwen 3.6, whatever, and we can access all those things from your website, which means that whether you are a thousand feet on the ground, you are flying in the air, you're in a rocket ship to space, you will be able to use your Hermes agent anywhere with no limitations, 100% private cuz it is it only exists on your physical laptop. That is the That is the private and local aspect to it. And as you can see, it's come down and it's given us some different ideas like Quant 32. B's really powerful. That's 32 billion parameters. And these models are just getting better and better. And obviously, because it's local and private, you will compromise a little bit on performance and maybe speed because you're not on a huge server. But that's effectively what people mean when they say run local and run private. Now, we've covered this one. It leads very nicely onto concept number eight, which essentially is never starting from zero. And so, I want you to imagine Hermes as a friend that never forgets what you tell her. And it will remember things like your birthday. And it could tell you things that you spoke about on specific days. This is what we mean when we talk about Hermes' memory. And so, for example, I could say something like, \"Hey there, man. Give me an example of a question I asked you on May 19th.\" Okay, so I can ask actually ask it questions about specific days and it can recall that and let me know based on the way that this memory works. And as you can see, the question I asked it was something about a famous thing about being liked and a really cool interesting quote about throwing rocks at enemies. So, it has within it the Hermes agent does a couple of different memory systems. It has a memory.md. And by the way, MD is just a file type. So, you know how it's like .pptx like PowerPoint or .whatever Word is .x. .md is just markdown file. All it literally is is just text with no special characters. Like that's basically it. It's plain, it's readable, and it lives here on your laptop. Then you've got an SQLite full text search of every message and every session. So, this effectively it can recall everything you've spoken about. Don't need to worry too much about the complicated mumbo jumbo, but just know that it's checking everything. And then the idea is that Hermes agent will compound quietly. So, ideally, if you you need to use it about 2 weeks for it to start feeling magical. And of course, you can supercharge this and connect it to other things like Obsidian memory systems. So, for example, in the Hermes agentic operating system I have, I've actually connected it to my Obsidian system here, which means that all of my Obsidian memory, and I'll put a link on screen if you haven't learned about Obsidian or Pancourt systems, you can check that out. And it also is connected to all of my Claude memory. So, anything I ever talked to Claude Code about, or any of the apps that I use on my computer, I can actually access that and it will dream for me overnight back proactively with suggestions based on questions that I've asked it or how I'm using it or what I'm spending my transcriptions, and it pulls together all these different you get information sources together to give you really specific stuff. To take its memory to a completely interstellar level because if your Hermes agent isn't getting information from stuff you're doing on your computer, Claude Code, Codex, how can it possibly give you the correct and accurate advice? Which is why I think the everything system, bringing it all together into one beautiful operating system, is really, really important, absolutely crucial. And this brings on to the beautiful concept number nine, which is what we call the character Bible, also known as the soul.md. So, this is the file that turns your chatbot into your personal assistant. It's somebody who gets your vibe, your communication style. If you're somebody who says, \"Hey dude, hey bro.\" or however you want to talk and communicate, it specifically explains that to the Hermes agent. And what you want to do basically is give it the information. For example, what is the time? What are your values? What is your style? How do you want them to communicate? And what I'm going to do for you down below is put a full soul.md template that I pulled together based on so many hours of using this stuff. It covers stuff like your identity, mission goals, you know, things about your business, how you want it to address. You want to give that to Hermes and think of it as like the character. Like, do you want to speak to like an honest watching type of character or like a Knight Rider type of character? You can give these soul.mds, and this is what makes your Hermes agent different from everybody else's is because it has a unique soul. Unless we specify what that is, how's it going to know what that looks like? So, super important, give it a soul, and make it fantastical. And again, if you're using an operating system, you can actually also see what those souls are in your own database. Now, concept 10 is the idea that Hermes itself comes off the shelf with many different integrations already locked in that. And then, what we would basically going to do here, guys, is just give it the specific API key for that thing. That could be text to speech, you can connect your entire world to Hermes' agent, and think of Hermes as a Swiss Army knife. For example, if I come back over here, and I open up my Hermes agent, I might say something like, \"Hey there, dude. What was the title of the last meeting I was in, and what is one action that I said I was going to take away?\" So, for example, I've connected it to Granola. Not a sponsor by any or anything like that, but basically, I use this to capture information on my meetings. It sits in there, and basically, now I can chat to this because I've configured it within my Hermes agent, and it will now allow me to ask any questions about anything. I could say, \"Hey, dude. What did I talk about 6 weeks ago for this thing to happen?\" Now, as you can see, it's going to come down here and ask my permission to run it. So, we're going to say, \"Yeah, that's absolutely fine. We're going to allow it for the session.\" And what's really cool here, guys, is actually come back and said, \"Hey, your your connection's expired,\" which is fantastic. I'm going to show exactly how you can add this in for any connection within Hermes cuz this is a great teaching moment. Now, check this out. \"Hey, dude. I would like you to give me a terminal command so that I can add my Granola API to your environmental variables.\" Oh my gosh, Jack, what does that mean? Well, effectively, what we never want to do is publish our API keys in the chat with Hermes agent. Because, again, everything is indexed and saved. So, what we do, especially if it's running on our computer, is we save it somewhere on the computer that it can access, but never actually sits in that. Now, check this one out. Let's come down here. All I'm going to do is copy this right now. I'm going to open up the terminal like so. All right, here's the terminal. Let me exit out of this process. Let me create a brand new one. Okay, beautiful. And what I'm going to do is just paste this one down here. And all it's going to ask me to do is change the API key, which is going to be right here. Then I come over to my app. I click on API keys. And then once you've created it, I've got this even simplified version here. And then you just literally replace this information here with your API key. And you're going to copy that, come to the terminal, and paste it in. That's done. I can say, \"Awesome. I've updated that. Please go ahead and try it again.\" And I wanted to include that in just so you get a sense of how you securely and privately update these, again, API keys within your system. Giving Hermes the ability to connect to anything. Anything that has an API key, you can connect to Hermes in exactly the same way. And what's cool about this is Granola also has an OAuth. So, if you just give it the name of the Granola MCP like this, it will actually ask you to authenticate and sign in. And look at this, guys. Hermes Agent would like to access your account. And the Hermes Agent actually opened up this MCP MCP server for me. So, I come down and allow access. And you can see authorization successful. You can now return to the CLI. And all I did, guys, honestly, I just came down. I copied this page. I just sent the link and shared it straight to Hermes. And as you can see, it's come down here and it's confirmed that it's Farm Blast Meeting, which is a community call I had, and exactly one of the questions that we asked. And that's And it's such a wonderful example of how you can connect your Hermes agent to your entire world. So, I could be anywhere and ask it any question I like. My concept 11 is the fact that Hermes lives on your computer. If you've set it up this way. Which means that not only can it do things like create files on your desktop or organize things on your desktop, it can also take actions on your computer. So, for example, it can do what? It can do browser vision, bash, fuse. What the hell does that mean? It has native Chrome DevTools, visual screen reading, real cursor movements, not a screenshot bar, an actual agency. So, if you're running to the gym and you're like, \"Damn, I completely forgot this thing.\" Basically, Hermes Agent can go ahead and do that. Now, to be fair, this isn't a feature I tend to use a lot, but again, if you just think about the fact that the agent lives on a computer, it can take actions and do various different things for you once you enable it with the tools. Concept 12 is around about using MCPs. MCPs stand for model contacts protocol. So, think of it like this, okay? You have an an API is like the wiring, and the MCP is like the instruction manual around that wiring. So, instead of having like a like, you know, a remote, for example, with a lot of exposed wires, MCPs are basically like instruction packages. It's like a universal remote control that lets you connect Hermes to effectively anything. We just used the Granola MCP to connect to basically my meetings. And we can do the same thing, for example, with Zapier MCP. So, Zapier MCP lets you spin up these servers like so, new MCP server, and I can go ahead and I can connect to anything that I wanted, right? If I come down here, for example, let's say that I want to come down and do other, I come over here and I connect to things that, you know, through this you can't connect to other applications. I can add tools. So, for example, I wanted to give it the ability to read emails, right? Well, how would I do that? Well, we can just basically select all the stuff that you want it to do. The one thing I wouldn't give it access to is the ability to send emails just yet. I would just hold off on that and its ability to reply. And when that's done, I can add these 10 tools. I can add in things like Google Drive, I can add in things like whatever I want to, Zapier. It's really cool. Then, all I do, I've got skills over here as a second tab. But, what we're really going to want to focus on here is the connection. And basically, what we can do now is generate a token here. And I've got an MCP server to connect to Zapier, which is awesome. Now, what I can do is come back over to Hermes, and I can say, \"Hey there, I'd like to connect to the Zapier MCP server. I've got my tokens. Could you give me a command to run through the terminal so I can add these to your environmental variables?\" Again, you can paste it, but I I'd often recommend that you do this in the terminal so it doesn't get recorded in your chat logs. And once you've done that, you can say, \"Awesome, I've done. Could you tell me the title of the last email I received?\" Again, then it will use an MCP to go ahead and do it. Now, why would you create an MCP versus just using an API key? It is way more what we call token efficient because all the checking it needs to do builds up the cost. An MCP basically explains exactly how it works so AIs can basically just look at the MCP manual and they'll know exactly all the stuff they can do, all the buttons they can press in that individual application. So, it knows how to draft emails, create emails. It has full understanding because it standardizes the use of that incredible tool for the agent. And just like that, we've got all the information that we need from this MCP. Concept 13 is understanding muscle memory for the agent. The idea here is that the longer we use Hermes agent, the better it gets. What I mean by that is consistently give it feedback as you go so it can iterate and improve as you're doing it. It has a series of Hermes skills it can pick up the more that you use it. It has various other things it can do and you can even if you want to build out different skills quite easily. So, for example, you can go to this website here, which is agentskills.ai. This gives you a good overview of what skills are if you want to go deeper on that. One of the things that I did actually guys is I built out something called the Pantheon, which effectively enables me, say for example, I want to add a persona, right? I can choose let's go with the alchemist for instance. I can explain its job. I can explain its description, give it a system prompt and then actually choose the model that I want to do it with because again, if it's like very intellectualized, we don't want Albert Einstein, you know, washing our car for example. You know, we want him working on the most impactful tasks. And so, you can build out skills like this. Look, I have a deep reasoning task I use for Claude, but you know, for autopilot jobs, I might use a lower cost model. So, you can build out these skills which give it the ability to do something different. So, for example, I could say something like, \"Hey there dude, I'd like to do some research for me, some very deep research on the best country in the world to live. Um you know, spin up a sub agent. Why don't you go ahead and use Deep Seek V4 and ChatGPT 2 critically analyze it in a nice little loop with Opus 4.7 reviewing everything. Um, just let's go ahead, use a cheap powerful model to go ahead and do loads of deep research, take some time with it, and then come back and share the results based on everything you know about me. Now, what's cool about this is it can delegate to different models to do different things for me, and it can use that as a skill. Now, it will use here for me the deep research skill that I've already created, which is fantastic and it just enables you to get so much more out of Hermes. Now, if anytime you want to stop it from doing a task, you can do {forward slash} stop, and that will actually basically stop Hermes from doing the thing. The concept 14 is about six keys to make it so unbelievably powerful. So, here we've got we've got Q. Now, what's interesting thing about Q here, okay? Now, Q is this, which means that let's say I ask it a question like, \"Hey dude, find out for me the best country for octogenarians. Which country world has the highest number of octogenarians?\" Okay? Really cool. I can send it off. Then let's say I want to do a follow-up task, I can do {forward slash} Q. And what this will do is send a prompt on the next turn so it doesn't interrupt. I can also do {forward slash} background, okay? And what {forward slash} background lets me do is effectively ask it let it do different tasks in the background. So, for example, I could come down and do background, and then I might come back and say something like, \"Hey, find out which country in the world has the healthiest 35-year-olds or 32-year-olds?\" Okay? Now, this will run in the background. Cuz normally when you use Hermes agent, basically what's happening is it's it's interrupting. So, this enables us in one chat window to have met multiple streams of thoughts running concurrently by using those two tasks. And as you can see, background task has gone ahead and actually physically got started. You got Kanban, which is for a a task board functionality you can have. Reset will clear it all up. Compress will compress the memory. So, important thing to bear in mind on the memory is that there is a certain size of the window. So, if you have, say for example, 10,000 words, and you ask it a question, every time you ask that chat agent a question, it is literally using all 10,000 words plus your question. And the longer it gets, basically, the lower its performance gets. So, what I do is one session, one window done, you want to reset. So, compressing basically means, \"Dude, could you just do me a favor and just summarize everything we talked about?\" So, it goes from this to like this, basically, and then you can continue the conversation. And as I showed you earlier, we can switch the models super easily. And what we can go down here and do {forward slash} model, and then we can basically Now we're talking to Opus, we're talking to ChatGPT, we can talk to anyone that we physically like. And I put a lot of the interesting ones down here that I thought were really quite interesting that you may find yourself using, you know, basically depending upon what situation you're in the moment. Now, concept 15 is what we call, basically, safety and security for the agent. These are house rules. Now, these agents have access to many different things, and because of that, I think it was Peter Parker told us this, right? With great power comes great responsibility. Might have been Ben's uncle. >> Remember, with great power comes great responsibility. >> But the point here is we basically don't want to give it the keys to the kingdom because that's responsible years, which is why I don't let it have the ability to send the emails to you. A lot of my buddies have done that, they've had their hands burned in the process, and it will get to the point where it will have less errors than humans. But at the moment, we want to follow what's called the principle of least access. And what that effectively means is Herme Herme's and any agent or system or anything you're building ever. I do this with all my clients. You only ever give it the minimal amount of access that it actually needs. The very, very minimal thing. So, for example, there's certain things that we don't want it to go ahead and do. API keys, never, ever, ever post into any chat like a Herme's especially specifically if it's keeping it. It's not that Herme's is bad person, it's just that if you're backing up your Herme's on a daily basis, we just don't want these keys. It's just one less thing for you to think about. So, if you can actually think about using it responsibly like that and follow this principle of this access, you will generally across all of your AI systems be extremely safe. Now, next thing to think about here in concept 16 is called the North Star and this is what's called the goal function. So, think about like this, okay? Rather than going question answer question answer, the idea of the goal functionality within Hermes is that you give it a North Star for a short-term project. So, I could say, \"Hey, I want to create a YouTube thumbnail that is pretty freaking epic.\" So, for example, I I come over to Hermes, right? And I come down here and why don't I just go ahead and clear this entire session. And so, for example, you come to the Hermes agent and you can give it a prompt like this, which is goal. I would like to decide whether I should take an opportunity A or B. Should I do a big consulting role at Anthropic or do I become head of innovation at OpenAI? I would like it to help me out. Help me make the best decision. And look at this. Now, it's come and said, \"Look, I'm going to assign a 20 goal budget as well a 20 turn budget to this.\" And effectively, what will happen now is our entire conversation exists to hit and crush that specific goal. So, it will ask me questions about what's about, what's going on, and it will get there on a mid-term basis. Now, what's really cool here is Hermes will not stop until it hits that decision. But, it does run into one limitation and that's the fact that not all goals we do in life can be solved in 20 turns and sometimes we've got to do things. This is why I built this idea of super goals and I'll put a link on screen if you want to look into this. The idea of super goals is that you explain the goal you want to, use the super goal skill, and effectively, what happens here is Hermes comes and asks you loads of questions and the output of that is like say four to 10 actions. Some that you do, some that Hermes does. For example, on this one, which is to launch a Hermes course and get 500 sign ups, one is shape the course promise, build an opt-in engine, write course sequence. Then, it's my turn. I have to go ahead and record the video promo and it lets me mark this task as complete when it's done and then my progress bar goes up. But, if I were to remove this and start from scratch, you'll see right now, look, every hearing is a great goal. I can copy this prompt. I can shoot straight back over to Hermes. What I can do is {slash} and I just come down here and click on stop. As if for instance, say gold pause is what we want to do. And the command I believe is goal clear. This will end the goal. And what I can do is come down, give it this information here, and in effect from now what this does will allow me to have a medium-term goal. And I'll say, \"Hey Jack, what is the goal that you want to go ahead and set?\" And so say, \"Hey dude, I would like to launch a paid course and get 1,000 uh email sign-ups to my new Hermes masterclass.\" Or whatever the thing is that you want to sell or or help people with. And then it starts to ask questions. I'll say, \"Hey, in 4 weeks, okay?\" And then it asks you questions dynamically. And I tested this so much, which is why I built all these different specific rules into it. And it's all connected to the dashboard. So when this is done, it will literally load up in the dashboard dynamically. And again, ask you different questions to get context. Cuz without context, how can it know that it's helping you out? Then all you ever do is you land on your mission control, you command it in Hermes, or if you wanted to, you can come down and use Claude. And it's the exact same thing, so you can pick up whenever you want to. Honestly, these mid-term goals, these chief working loops, are so freaking powerful. You can plan your life, hit these mid-term goals using Hermes agent. But concept 17 is about using Hermes agent as a team. So the idea is that we can tag in different models to various different things. You can spin up what is called sub agents. So for example, if I come back over now to Hermes agent, what I might do is just say, \"Hey, I'm just going to pause that for a second. What I would like to do is go ahead and research for me the best AI companies to work for on the planet. I'd like you to spin up two sub agents. One is going to go ahead and look at the ones in America, and the other one is going to look at alternative jobs maybe internationally, okay?\" And what Hermes can do is spin up sub agents. And what sub agents are is think about like this. You're at Hermes, and then underneath Hermes you have two other agents, each with their own fresh contacts. And it delegates these tasks, these two agents go ahead, they do all the work, and they report back. Now, this is happening in parallel, not sequentially, which means that I could have four, five, six different agents all researching things like a team doing work individually and at the end of the day or the end of the turn, everyone hands in their paper, their research, and instead of waiting, you know, six hours, it's done in one hour instead. This is called parallelization, parallel agents, and gets you significantly further faster using Hermes. I use this strategy all the time to get the most out of the agents. For example, here, Hermes co-founder Teknum, 12 parallel Hermes agent enters. I hope I pronounce his name correctly there. Every day to build Hermes itself, monitoring issues, dog food in the Kanban good board, parallel sub-agents are the power user move. So, even the guy that built Hermes agent used parallel agents to do it. So, again, multiple agents working together in tandem to one objective. Or it could be individual objectives, right? One could be doing research, one could be writing, one designer, one scheduler, and they all put these beautiful folders inside your desk. The concept 18 is so powerful. This was called the heartbeat or the cron job. Cron job sounds so freaking complicated. One of the coolest things you can do is this idea that it keeps working whilst you sleep, the 24/7 AAA agent, okay? A heartbeat effectively is something that you set up to happen every single day. It's essentially called a zombie as well, a zombie process, if you like. And effectively, what it will do is you give it something to do. So, effectively, a heartbeat is silent until something dies. So, a sub-agent pings every few seconds, zombie detection, fresh agent reclaimed to task. Per task retries with hallucination recovery. What does that mean? Let me show you what a cron is first, and then I think it'll make a little bit more sense. So, a cron agent here, if I come down here and just say, \"Hey, dude, I want you to do two things for me. I want you to send me a reminder in 30 seconds that the sky is blue, okay?\" This is a scheduled task. This is basically in a certain amount of time, it is going to come and give me an update. Now, for example, I have a daily morning briefing, okay? And I say, \"I want you to go through my um, email list, I want you to go through my calendar, I want you to go through everything I'm doing that day, use all the information that you know about me, and give me, you know, a daily morning brief of, say, I don't know, five things that are relevant. And it can build that for you. So, every day you get that on repeat, basically. So, you can schedule these wonderful things inside Hermie's Agent, get to come to you at 8:00 a.m. every single morning. But, the one that you're going to find yourself using most of the time are these scheduled tasks that make it feel like it's alive because proactively coming to you and asking questions. You could say, \"Hey, uh, periodic times randomly come and ask me questions to check in on me, do a lot of other things.\" Then, we go to concept number 19, which is essentially the dirty secret not a lot of people want to talk about, which is, can you believe, 73% of every quest is fixed overhead. So, what we're talking about here is budget, okay? Now, what I mean by that is that, like, if you are using API keys, you can very easily burn through a lot of money, which is why we want to be very specific about the specific models we're using. Now, tokens, you know, um, a parking meter takes maybe dollar coins, um, AI models take tokens. Roughly speaking, it a token, I guess a 10 tokens is about seven words. It's about seven seven 70 75%. That's effectively what the translation looks like. So, just make sure that you often clear the session and that you're using the right model for the right job when you're using Hermie's to keep your token cost usage all the way down. And there's loads of mini strategies, like, don't overload it with a bazillion skills that you don't need, don't make it system prompts too long, keep it nice and clean, and one conversation, one goal, then it's always best practice to clear, and refresh, and start again. And just point it out, some guy actually spent 4 million tokens in 2 hours of light usage, 21,000 tokens, just asking about the weather because he made a mistake. So, it just gives you a good a good idea of how you can reduce those down. Now, the 20th concept here is this idea that it isn't just a chatbot, Hermie's Agent can be part of an operating system. Well, we have various different personas in the Pantheon. We can have a memory vault that extends its memory capabilities to a significant level, basically enabling you to call upon any email or meeting that you've ever had, any knowledge source from any YouTuber you've ever seen, connected to notebook 11, a gargantillion different things. We can have, for example, so many different tools and effectively when we create an operating system, it just gives it a place to manage everything. I call this a kind of everything of AI. So, if you look at this Claude code operating system, we have our AI spendage, how much money we're making from our skills, we can see our usage in real time, dreaming improvements. Like I said, they will come to you proactively, mission control, our skills and memory systems. Now, in the Hermes operating system here, I can see at a glance all of the connections it has available to it. So, I know exactly what it's connected to. I know what model I'm using, I know my memory, I know my usage. I can chat to it directly. I have this ability to manage goals and plans with Hermes agent. So, whatever I'm working on, I can work with Hermes agent in tandem on that. I have this Pantheon, so I can create these beautiful I don't know I personally like to visually see my graphics and what I'm dealing with. So, I can like, well, actually now I really want to kind of upgrade that like I'm missing this or I'm missing that. I can grab different things and build various different things. I just think it's a nicer user interface to be able to go ahead and build things like that. Again, and what we can do here crucially and I highly recommend you do is connect your Hermes to GitHub. Now, GitHub is a basically code for place where we store stuff. Okay? It's like the world's repository. It's basically you can store information and files. And what you can do with Hermes agent, guys, is essentially have your entire Hermes agent backed up on GitHub every single day. Mine does that. And what that means is if you ever change your computer or God forbid you lose it or something, you have that backed up somewhere, right? We can connect to GitHub. We've got all the profile information here. We can connect to our Obsidian brain, which is this beautiful memory system here that basically just lets us see everything we're doing with different memory systems. We've got there's my Claude and it just enables you to get a level of control. And crucially it comes on to the 21st concept which is the idea of one brand. The idea that we are going to connect the Hermes agent, the business brand, with Claude code. We talked about the differences but here's the difference guys. If we can actually connect them together, it unlocks incredible capabilities. Hermes knows your projects, your customers, your decisions. It lives everywhere. everywhere. Maybe it's like cotton who knows what it is. And then you got Claude code is this precision tool that we use for building. And actually what happens is when we can physically connect these ones together, we have this shared memory and understanding where we're not operating in isolation. It's like you say one thing to Hermes agent, another to Claude code. How do you know they connect together in a meaningful way unless they have access to basically what they're doing at different times. And so now we've covered these 21 core concepts in plain English. The next thing we need to do is leverage all of these and get one of these operating systems ourselves which you can learn about in this video right here.",
  "timestamped_text": "0:00 Hermes is the world's most powerful AI\n0:02 agent, but if you're new, it can be\n0:04 incredibly overwhelming. You're probably\n0:07 looking at the app and asking, \"How does\n0:09 this make my life easier? What is a\n0:11 habit? What is an operating system? How\n0:13 does this work?\" And without knowing the\n0:15 basics, you will never unlock its full\n0:17 capabilities. So, in this video, I'm\n0:19 going to give you the exact tutorial I\n0:22 wish I had when I first started. We'll\n0:24 break down 21 Hermes concepts, starting\n0:27 from extremely simple building to its\n0:30 most powerful features. And by the end\n0:32 of the video, you'll understand this\n0:34 incredibly powerful technology, even if\n0:37 you're a complete beginner. And if\n0:38 you're new, I'm Jack. I built and sold\n0:40 my last tech startup with a gazillion\n0:41 customers. Now, I'm building my own AI\n0:43 startups, and I share here the stuff\n0:45 that actually works. So, if you haven't\n0:47 already, grab that beautiful coffee, and\n0:50 let's dive straight in. So, let's begin.\n0:51 Every Hermes concept explained for\n0:54 normal people. Now, this is going to be\n0:55 the Hermes tutorial. If you just watch\n0:57 this video, you will be able to walk\n0:59 into any room, and people will think,\n1:01 \"Dude, who the hell is this guy? Who is\n1:03 this Hermes genius? Does he work for\n1:05 Hermes?\" That's what you They're going\n1:06 to feel at the end of the video. Now,\n1:09 Hermes is massive. 271 billion tokens\n1:14 have used, which is around about 240\n1:16 billion, 230 billion words. It's across\n1:18 22 messaging platforms. It is huge.\n1:21 First concept is the foundation, which\n1:23 is basically it is what we call an\n1:25 agent, not a chatbot. This is\n1:27 fundamental to how you understand and\n1:30 think about the Hermes agent. A chatbot\n1:32 tells you how to book a flight. An agent\n1:36 will actually go ahead and book the\n1:37 flight and find it for you. So, think\n1:39 about this. With a chatbot like ChatGPT\n1:41 and Claude, you give it a goal, and it\n1:43 creates a plan. But with an agent, it\n1:45 has tools, and it can actually go ahead\n1:47 and do actions for you. So, for example,\n1:50 what a lot of people don't really\n1:51 understand here, if you think about it\n1:52 like this, right? It's like a smart\n1:53 friend versus is personal assistant.\n1:56 Hermes is a personal assistant. Let me\n1:59 show you what I mean. So, here for\n2:00 example, I have the Hermes agent\n2:01 connected. If you haven't set it up and\n2:03 you want to learn that, I'll put a link\n2:04 on screen to check out. I can say, \"Hey\n2:06 there, my man. I'd like to find for me\n2:08 the cheapest flight from Dubai to\n2:10 Toronto. Flying out, let's say, at any\n2:13 point within the next 2 weeks. Let me\n2:14 know what that looks like.\" You can ask\n2:16 it questions and this will go ahead and\n2:18 use its tools, its agentic capabilities\n2:21 to take actions for me, find\n2:23 information, and enables me to do\n2:25 anything that I like with it. And so,\n2:26 it's done a manual check. I then said,\n2:28 \"Find me one flight and open it up for\n2:29 me in the HTML.\" It looks gorgeous. It\n2:31 went ahead and did that and as you can\n2:32 see, I now have it there. So,\n2:34 conceptually, Hermes is an AI with\n2:37 access to tools. Could be your Gmail,\n2:39 your calendar, many different things\n2:41 that makes it so powerful. So, wherever\n2:43 you are, wherever your phone lives, you\n2:45 can enact things, you can affect change\n2:47 with your AI equipped with these tools.\n2:49 The second concept of Hermes is what it\n2:51 actually is and when you would use it\n2:53 versus using Cloud Code, ChatGPT, Codex,\n2:57 Grok, all these different things. I want\n2:59 you to think of them in this as a\n3:00 framework to understand where Hermes\n3:02 fits in with this. So, think about this.\n3:03 Hermes is like your beautiful Labrador.\n3:05 It's your dog. Cloud Code is your\n3:06 contractor. Um OpenFlow itself is your\n3:09 roommate. And then, Antigravity can be\n3:11 like your your IDE. It's your it's your\n3:13 buddy, the guy that you work with. The\n3:14 best way to think about this is if\n3:16 Hermes is your dog, it lives with you.\n3:18 It knows you. It gets smarter year over\n3:21 year. And Cloud Code is a contractor.\n3:24 It's brilliant for a specific job, but\n3:26 it doesn't necessarily remember you. The\n3:28 core thing about Hermes and the whole\n3:30 point of it is that the more that you\n3:31 use it, the more it gets to know you,\n3:33 the better it gets, and the more\n3:35 specific it is with you. And what it\n3:37 basically means is you can use it\n3:39 wherever you go, wherever your your\n3:40 mobile, you're in a coffee shop, a gym,\n3:42 you can whip out your phone, chat to an\n3:44 agent whose memory is always always\n3:46 improving about you. Now, can I have,\n3:48 for example, Cloud Code learn stuff\n3:50 about me, too? Of course, you can. We\n3:52 can get those memory systems. But I did\n3:54 with Hermes is we can take it anywhere\n3:55 where there's an it enables us to be\n3:57 fully mobile, fully custom, and just\n3:59 unlock some incredible capabilities. So,\n4:02 if I'm at the desktop, I might be using\n4:04 Claude, but if I'm out and about, I'm\n4:05 going to be using Hermes because the\n4:07 more you use it, the better it\n4:08 understands you. And its whole design\n4:11 ethos it's built to live with you and\n4:13 improve with you over time. Which brings\n4:14 us on to, importantly, the third\n4:16 concept, which is one brand, 22 mouths.\n4:19 What do I mean about that? Hermes itself\n4:22 can be accessed from over 22 different\n4:25 interfaces. We use Telegram, but for\n4:27 example, we could talk to it on Discord.\n4:30 We could talk to it on WhatsApp. We\n4:31 could talk to it on any different\n4:33 interface. You can even talk to it on\n4:35 different operating systems. For\n4:36 example, this is my Hermes operating\n4:38 system that connects to my Claude code\n4:40 operating system. And I can ask it\n4:41 questions. I'm like, \"Hey there, man.\n4:43 What day is it today?\" And I'm just\n4:44 speaking directly to my Hermes agent. I\n4:46 can see every single chat I've had on\n4:48 the left-hand side. And with Hermes, you\n4:51 can actually access the same brand, the\n4:53 same intelligence from all of these\n4:55 different locations. You aren't just\n4:57 constrained to one single input because\n5:00 where the Hermes is running, whether on\n5:02 your computer or VPS, it's the same\n5:04 intelligence and we can connect to it\n5:06 all these different interfaces. For\n5:08 example, if you come and use the Hermes\n5:10 setup code, for example, you'll be able\n5:12 to see all of the different locations\n5:14 that you can actually use with Hermes.\n5:15 For example, this shows you all the\n5:16 different message platforms Slack,\n5:18 Matrix, and all you can do is scroll\n5:20 through these, select them, and then you\n5:21 can connect and chat to the same AI\n5:23 intelligence from any platform that you\n5:25 like. You aren't locked in to just one\n5:27 software. And even in this one here, for\n5:29 example, if you download this Claude\n5:30 code operating system Hermes operating\n5:32 system, you can chat to it in your\n5:34 browser, which I think is a bit more of\n5:35 a dynamic interface. And so, if you\n5:37 think about it as as an analogy, that\n5:39 imagine that you have an operator\n5:41 sitting in the middle, and you can call\n5:42 into this operator from any different\n5:44 one of these interfaces, and she will\n5:46 route every single one of them to you\n5:47 guessed it, your Hermes agent. Which\n5:50 takes us critically onto the next one,\n5:52 which is going to be constant number\n5:53 four, which is where Hermes lives. Now,\n5:56 you've probably heard of it a few\n5:57 different terms like the terminal and\n5:59 VPS and locally hosted. And this is\n6:02 what's really important to understand.\n6:03 The first one is your own computer,\n6:06 which you can run completely for free.\n6:08 This is the easiest, it is the quickest,\n6:10 and in some respects, it's actually\n6:12 incredibly safe to do it this way rather\n6:14 than on other options that I'll show you\n6:16 in a second. You don't need to go and\n6:17 buy some crazy host somewhere. You can\n6:19 do it completely for free on your\n6:20 computer. To do that, all you literally\n6:22 need to do is come over to the Hermes\n6:23 website or click the link below. You\n6:25 copy this, you open up the terminal, and\n6:27 the terminal is basically the way you\n6:30 talk to your computer. It accepts code.\n6:33 And to pull it up, you do command and\n6:34 space bar, you type in terminal, and the\n6:36 terminal will appear. And all you do is\n6:38 control V that command, and effectively,\n6:41 that will install Hermes into your\n6:42 computer. Alternatively, you can go to\n6:45 an app like Cloud Code, and you can say,\n6:47 \"Hey there, I'd like you to install\n6:48 Hermes, please. This is the install\n6:50 code. Go and grab it from the GitHub.\"\n6:52 Or whatever you want to, right? And you\n6:54 can paste it in like that. And\n6:55 literally, Cloud Code will install it\n6:57 for you. So, this is what we mean when\n6:59 we talk about the terminal. Now, it\n7:01 running on your computer just means that\n7:03 as long as your computer is on, Hermes\n7:05 will run. And this is why lots of people\n7:07 have got either like, you know, old\n7:09 MacBooks that they run it on or or\n7:11 MacBook desktop or different kind of\n7:13 laptop, and it just runs for you 24/7 on\n7:17 that computer, or it can just exist on\n7:19 your computer. But effectively, if the\n7:21 laptop shuts down, Hermes doesn't work.\n7:24 So, that's effectively this local\n7:25 hosting revolution. I'm a big believer\n7:28 in local hosting. It's here. It's in my\n7:30 room. It's It's here. I can I can touch\n7:32 it. I can feel it. The second one is a\n7:34 VPS. So, this is essentially where\n7:36 instead of running on your computer,\n7:38 it's actually running on someone else's\n7:40 computer somewhere else in the world.\n7:42 And these companies called hosting\n7:44 providers effectively say, \"Hey, what\n7:46 we're going to do is buy like thousands\n7:48 of these like, you know, computers like\n7:50 this, and we're going to stack them up\n7:51 or whatever they're doing over there,\n7:53 and you're basically renting a virtual\n7:56 private server.\" Okay, so it's running\n7:58 over there. But one of the risks with\n8:00 this is the fact that it it's it's\n8:01 running somewhere else. They can be\n8:03 attacked. You need to do certain things\n8:04 to certify and protect the ports, and\n8:07 you pay monthly for renting that\n8:10 service. For example, our speech attacks\n8:12 at Glyder, that lets me say things and\n8:14 speak as if for instance, you know, we\n8:16 ran some space and for inference to make\n8:18 that fast. So, VPS is an interesting\n8:20 one, and you'll pay monthly for that.\n8:21 And that's what they mean when they say\n8:22 virtual private server. It's running\n8:25 remotely somewhere else. And so, just be\n8:26 aware that these virtual private server\n8:28 companies do pay people money if you\n8:31 sign up using links. So, it's not always\n8:33 necessarily the best thing. Now, this is\n8:35 run here on your computer and free. So,\n8:37 for example, if I come over here, I can\n8:38 say, \"Hey there, I'd like to shut down\n8:41 my Hermes agent, please.\" And you can\n8:43 literally chat to Claude to basically do\n8:45 different things. Now, you'll see as it\n8:47 does that, you'll see my Hermes button\n8:48 now will come down and say, \"Gateway is\n8:50 shutting down\" as this process goes down\n8:53 and closes for us. And as you can see,\n8:54 just like that, Hermes says, \"Gateway\n8:56 shutting down. Your current task will be\n8:57 interrupted.\" And it shut down, and I\n8:59 can say, \"Awesome. Go ahead and reboot\n9:01 him up for me, please.\" And you can do\n9:02 the same thing. And this can happen\n9:04 because it's running on my own actual\n9:06 computer, which is on 24/7. And this\n9:08 leads us on to the next and fifth\n9:10 concept, which is OAuth versus API. What\n9:14 the hell does that mean? So, there's two\n9:16 ways that you can connect to something\n9:17 essentially. One is what we call open\n9:19 authentication. It is dead straight or\n9:21 forward, and all it basically does is\n9:23 opens up a new browser, and you sign in.\n9:26 Let me show you exactly what I mean. So,\n9:28 here we go. We've got, for example, a\n9:29 terminal. So, we can chat to Hermes.\n9:31 Might come down on my dashboard on the\n9:33 left-hand side to have a look at the\n9:34 Hermes setup command. Let me just go\n9:36 ahead and copy this, throw this in the\n9:38 terminal, and you'll see what I'm in.\n9:39 So, as you can see, it's actually popped\n9:40 up and said Grok, Super Grok, and\n9:42 Premium. Now, for example, if you look\n9:44 at models, which come to shortly, or any\n9:46 servers, we could either do an OAuth.\n9:47 So, for example, if I come down here and\n9:49 I press enter, I might come down and\n9:51 reauthenticate. So, let me come down\n9:53 here and press number two, and you'll\n9:54 see what I'm in. I just want to show\n9:55 you, rather than explaining that. Look,\n9:57 this has opened up a new window. This is\n10:00 open authentication. So, instead of me\n10:02 giving it keys, it just says, \"Cool,\n10:03 well, you're signed in, right, bro?\" So,\n10:05 why don't you just go ahead and allow\n10:07 me, so you can literally come over here,\n10:09 you click on allow, and then this will\n10:10 now be fully connected to Hermes.\n10:12 Connection successful, and it is easy as\n10:15 that. Now, we're fully connected, and it\n10:16 lets you pick the model. How cool is\n10:18 that, right? Now, if we come and do that\n10:20 again, little hack, by the way, if you\n10:21 press up on terminal, it will actually\n10:23 show you the last thing that you\n10:24 messaged it, which is Hermes setup. And\n10:26 let's say we go now and do a different\n10:27 one. Let's say we want to do OpenRouter.\n10:29 I press spacebar to select it. Okay,\n10:31 come down. It's going to say active\n10:33 provider, and it's asking here for an\n10:35 OpenRouter API key. So, what this means\n10:37 is it needs a physical key. So, you come\n10:39 over to openrouter.ai/models,\n10:42 and from here, we're going to grab a\n10:43 key, and then input it basically here.\n10:46 So, those are the two ways you're ever\n10:47 going to connect. One is open\n10:49 authentication, and the second one is\n10:51 API key. Now, OAuth itself, you can use\n10:53 basically Grok, which is incredible, and\n10:56 you can use ChatGPT, but you cannot use\n10:59 Claude. With OAuth and API key, you can\n11:02 access all the different models, which\n11:03 I'll touch on very shortly. Best way to\n11:05 think about it is that an OAuth is like\n11:07 a sign-in. Okay, so one is a button that\n11:09 you can take back, and the other is a\n11:11 string of characters that live on a\n11:13 server somewhere, and anyone that use an\n11:15 API key with, you can just rotate it\n11:18 anytime you want to. And then once you\n11:19 rotate it in that environment, which is\n11:22 what OpenRouter is so good, it will\n11:23 never work again. So, that's like the\n11:25 big difference. And this takes it on\n11:26 very nicely onto phase two, and actually\n11:29 concept six, which is choosing the right\n11:31 model that that powers us. So, if you\n11:33 think of Hermes as a series of systems,\n11:36 okay? And I'll show you this down and\n11:37 show you exactly what I mean. Hermes\n11:38 itself, right, is an agentic framework\n11:41 with tools and capabilities. But,\n11:43 powering Hermes is a brain. So, for\n11:45 example, in Hermes, I can do {forward\n11:47 slash} type in model like so, and it\n11:49 will let me change the model, the brain\n11:52 behind Hermes. So, at the moment, it's\n11:54 using Grok, but I could switch to Open\n11:56 Router, which gives me access to all of\n11:58 the models on the planet. I can get to\n12:00 Anthropic, I can get to OpenAI Codex, I\n12:02 can do any of these things that I want\n12:04 to, GPT 5.5. Now, if you're using\n12:06 Hermes, you must connect your ChatGPT\n12:09 account to it. And the reason for that\n12:11 is you can use your $20 subscription to\n12:14 code. You cannot do that with Claude.\n12:17 You have to pay for API credits with\n12:18 Claude. So, you're going to see that GPT\n12:20 5.5 is like the one of the best ways you\n12:23 can use it there a lot using your\n12:24 existing subscription, and it's exactly\n12:26 the same with Grok, which is why it's so\n12:28 freaking cool. Now, the idea here is\n12:30 that we want to pick the right tool for\n12:32 the right job. I am model agnostic. I\n12:34 believe in the multi-brain strategy,\n12:37 which is basically model agnosticism, or\n12:39 we're agnostic. And effectively, we use\n12:41 the best model for the specific task.\n12:43 So, the best model at reasoning right\n12:45 now, we believe, is Opus 4.7. So, what I\n12:47 do when I want to reason is I come over,\n12:49 I change the model like so, okay? And I\n12:52 will just go ahead and chat to this on a\n12:53 daily basis. I click on Open Router, and\n12:56 I pick Opus 4.7. Now, I pay for credits\n12:59 on this, but in Open Router, when you\n13:00 create an API key, you can set I want\n13:03 this to be $10 maximum per month. That's\n13:06 it. And if it hits $10, it just stops\n13:09 working, and you can go ahead and switch\n13:10 the models, okay? Then you've got, for\n13:12 example, other models that can do\n13:14 volume. So, GPT is an amazing\n13:16 generalist. Grok, again, you can use for\n13:18 volume. Grok, when you authenticate with\n13:20 your X account, can also search Twitter,\n13:22 as well. And then you can bring in other\n13:24 models like Deep Seek, and actually run\n13:27 your entire thing completely for free\n13:30 using a series of free models if that's\n13:32 what you wanted to. So, I want you to\n13:33 think about your brain more as a toolbox\n13:36 rather than a one-size-fits-all because\n13:38 to a hammer everything is a nail, but\n13:40 sometimes you want to use different\n13:41 things like Claude for design and\n13:43 ChatGPT for reviewing code and being\n13:45 more general activities. So, we're just\n13:46 going to tag in the models that we want\n13:48 to and to do that we typically want to\n13:50 use something like OpenRouter because\n13:52 that's one connection that gives you\n13:54 hundreds of different models. little\n13:55 fact by the way, if you come over to\n13:57 openrouter.ai/rankings,\n13:59 you can see month on month by the way\n14:01 who where all where all the things are\n14:02 being used. Look at this. Deepseek V4\n14:04 Flash, right? That's roughly The Flash\n14:06 model is not as powerful as the main\n14:08 model, but effectively you can get like\n14:09 95% performance for 1% of the cost.\n14:12 That's kind of how cool these different\n14:14 systems are. And Concept Seven is a data\n14:16 of what we call local hosted. And local\n14:19 hosted just means that not only is\n14:20 Hermes running your laptop, but the\n14:22 actual model itself is running on your\n14:25 own computer. Now, the trade-off for\n14:26 this is that the biggest models are\n14:28 billions and billions of parameters,\n14:31 which means you need data centers that\n14:32 are huge to run them. So, you're going\n14:35 to be limited based on that, but what\n14:36 you can do on your MacBook, for example,\n14:37 is click the Apple icon at the top,\n14:39 click on about this Mac, and you can see\n14:41 some information about it. So, you can\n14:43 basically screenshot that Mac, you can\n14:44 push that image into Claude or Hermes\n14:46 and say, \"Hey, for example, if I come\n14:48 down here and do this real quick, I\n14:49 screenshot this, I share it to Hermes, I\n14:51 can paste the image in and be like,\n14:53 \"Hey, based on my MacBook\n14:54 specifications, what do you think would\n14:56 be the most powerful local model that I\n14:58 could run?\" Okay? And what we can do\n15:00 that is actually download that model\n15:02 using something called Ollama that lets\n15:04 you download stuff. And you can even do\n15:06 Ollama Cloud, which is a free model\n15:08 hosted somewhere, or I can download it\n15:10 specifically like Gemma 4, Qwen 3.6,\n15:13 whatever, and we can access all those\n15:14 things from your website, which means\n15:16 that whether you are a thousand feet on\n15:17 the ground, you are flying in the air,\n15:20 you're in a rocket ship to space, you\n15:22 will be able to use your Hermes agent\n15:24 anywhere with no limitations, 100%\n15:27 private cuz it is it only exists on your\n15:30 physical laptop. That is the That is the\n15:32 private and local aspect to it. And as\n15:34 you can see, it's come down and it's\n15:35 given us some different ideas like Quant\n15:37 32. B's really powerful. That's 32\n15:39 billion parameters. And these models are\n15:41 just getting better and better. And\n15:43 obviously, because it's local and\n15:45 private, you will compromise a little\n15:47 bit on performance and maybe speed\n15:49 because you're not on a huge server. But\n15:50 that's effectively what people mean when\n15:52 they say run local and run private. Now,\n15:55 we've covered this one. It leads very\n15:56 nicely onto concept number eight, which\n15:59 essentially is never starting from zero.\n16:02 And so, I want you to imagine Hermes as\n16:04 a friend that never forgets what you\n16:06 tell her. And it will remember things\n16:07 like your birthday. And it could tell\n16:08 you things that you spoke about on\n16:10 specific days. This is what we mean when\n16:13 we talk about Hermes' memory. And so,\n16:14 for example, I could say something like,\n16:16 \"Hey there, man. Give me an example of a\n16:18 question I asked you on May 19th.\" Okay,\n16:21 so I can ask actually ask it questions\n16:23 about specific days and it can recall\n16:25 that and let me know based on the way\n16:27 that this memory works. And as you can\n16:28 see, the question I asked it was\n16:29 something about a famous thing about\n16:31 being liked and a really cool\n16:33 interesting quote about throwing rocks\n16:34 at enemies. So, it has within it the\n16:36 Hermes agent does a couple of different\n16:38 memory systems. It has a memory.md. And\n16:40 by the way, MD is just a file type. So,\n16:43 you know how it's like .pptx like\n16:45 PowerPoint or .whatever Word is .x. .md\n16:49 is just markdown file. All it literally\n16:51 is is just text with no special\n16:53 characters. Like that's basically it.\n16:55 It's plain, it's readable, and it lives\n16:57 here on your laptop. Then you've got an\n16:58 SQLite full text search of every message\n17:01 and every session. So, this effectively\n17:03 it can recall everything you've spoken\n17:05 about. Don't need to worry too much\n17:06 about the complicated mumbo jumbo, but\n17:08 just know that it's checking everything.\n17:09 And then the idea is that Hermes agent\n17:11 will compound quietly. So, ideally, if\n17:13 you you need to use it about 2 weeks for\n17:14 it to start feeling magical. And of\n17:16 course, you can supercharge this and\n17:18 connect it to other things like Obsidian\n17:20 memory systems. So, for example, in the\n17:22 Hermes agentic operating system I have,\n17:24 I've actually connected it to my\n17:26 Obsidian system here, which means that\n17:28 all of my Obsidian memory, and I'll put\n17:30 a link on screen if you haven't learned\n17:31 about Obsidian or Pancourt systems, you\n17:33 can check that out. And it also is\n17:35 connected to all of my Claude memory.\n17:36 So, anything I ever talked to Claude\n17:38 Code about, or any of the apps that I\n17:41 use on my computer, I can actually\n17:42 access that and it will dream for me\n17:44 overnight back proactively with\n17:46 suggestions based on questions that I've\n17:48 asked it or how I'm using it or what I'm\n17:50 spending my transcriptions, and it pulls\n17:52 together all these different you get\n17:54 information sources together to give you\n17:56 really specific stuff. To take its\n17:57 memory to a completely interstellar\n18:00 level because if your Hermes agent isn't\n18:02 getting information from stuff you're\n18:03 doing on your computer, Claude Code,\n18:06 Codex, how can it possibly give you the\n18:08 correct and accurate advice? Which is\n18:10 why I think the everything system,\n18:12 bringing it all together into one\n18:13 beautiful operating system, is really,\n18:15 really important, absolutely crucial.\n18:17 And this brings on to the beautiful\n18:19 concept number nine, which is what we\n18:20 call the character Bible, also known as\n18:23 the soul.md. So, this is the file that\n18:25 turns your chatbot into your personal\n18:27 assistant. It's somebody who gets your\n18:29 vibe, your communication style. If\n18:31 you're somebody who says, \"Hey dude, hey\n18:33 bro.\" or however you want to talk and\n18:35 communicate, it specifically explains\n18:37 that to the Hermes agent. And what you\n18:39 want to do basically is give it the\n18:41 information. For example, what is the\n18:43 time? What are your values? What is your\n18:45 style? How do you want them to\n18:46 communicate? And what I'm going to do\n18:47 for you down below is put a full soul.md\n18:50 template that I pulled together based on\n18:51 so many hours of using this stuff. It\n18:53 covers stuff like your identity, mission\n18:55 goals, you know, things about your\n18:56 business, how you want it to address.\n18:58 You want to give that to Hermes and\n19:00 think of it as like the character. Like,\n19:02 do you want to speak to like an honest\n19:03 watching type of character or like a\n19:05 Knight Rider type of character? You can\n19:07 give these soul.mds, and this is what\n19:09 makes your Hermes agent different from\n19:11 everybody else's is because it has a\n19:13 unique soul. Unless we specify what that\n19:15 is, how's it going to know what that\n19:17 looks like? So, super important, give it\n19:19 a soul, and make it fantastical. And\n19:21 again, if you're using an operating\n19:21 system, you can actually also see what\n19:24 those souls are in your own database.\n19:26 Now, concept 10 is the idea that Hermes\n19:27 itself comes off the shelf with many\n19:29 different integrations already locked in\n19:32 that. And then, what we would basically\n19:34 going to do here, guys, is just give it\n19:35 the specific API key for that thing.\n19:38 That could be text to speech, you can\n19:40 connect\n19:43 your entire world to Hermes' agent, and\n19:46 think of Hermes as a Swiss Army knife.\n19:48 For example, if I come back over here,\n19:50 and I open up my Hermes agent, I might\n19:52 say something like, \"Hey there, dude.\n19:54 What was the title of the last meeting I\n19:56 was in, and what is one action that I\n19:58 said I was going to take away?\" So, for\n20:00 example, I've connected it to Granola.\n20:02 Not a sponsor by any or anything like\n20:03 that, but basically, I use this to\n20:05 capture information on my meetings. It\n20:07 sits in there, and basically, now I can\n20:09 chat to this because I've configured it\n20:11 within my Hermes agent, and it will now\n20:13 allow me to ask any questions about\n20:15 anything. I could say, \"Hey, dude. What\n20:16 did I talk about 6 weeks ago for this\n20:18 thing to happen?\" Now, as you can see,\n20:20 it's going to come down here and ask my\n20:21 permission to run it. So, we're going to\n20:22 say, \"Yeah, that's absolutely fine.\n20:24 We're going to allow it for the\n20:24 session.\" And what's really cool here,\n20:25 guys, is actually come back and said,\n20:27 \"Hey, your your connection's expired,\"\n20:29 which is fantastic. I'm going to show\n20:31 exactly how you can add this in for any\n20:32 connection within Hermes cuz this is a\n20:34 great teaching moment. Now, check this\n20:35 out. \"Hey, dude. I would like you to\n20:37 give me a terminal command so that I can\n20:40 add my Granola API to your environmental\n20:44 variables.\" Oh my gosh, Jack, what does\n20:46 that mean? Well, effectively, what we\n20:48 never want to do is publish our API keys\n20:51 in the chat with Hermes agent. Because,\n20:53 again, everything is indexed and saved.\n20:55 So, what we do, especially if it's\n20:56 running on our computer, is we save it\n20:58 somewhere on the computer that it can\n21:00 access, but never actually sits in that.\n21:02 Now, check this one out. Let's come down\n21:04 here. All I'm going to do is copy this\n21:06 right now. I'm going to open up the\n21:07 terminal like so. All right, here's the\n21:09 terminal. Let me exit out of this\n21:10 process. Let me create a brand new one.\n21:11 Okay, beautiful. And what I'm going to\n21:13 do is just paste this one down here. And\n21:15 all it's going to ask me to do is change\n21:16 the API key, which is going to be right\n21:18 here. Then I come over to my app. I\n21:20 click on API keys. And then once you've\n21:21 created it, I've got this even\n21:23 simplified version here. And then you\n21:24 just literally replace this information\n21:26 here with your API key. And you're going\n21:28 to copy that, come to the terminal, and\n21:30 paste it in. That's done. I can say,\n21:31 \"Awesome. I've updated that. Please go\n21:33 ahead and try it again.\" And I wanted to\n21:35 include that in just so you get a sense\n21:36 of how you securely and privately update\n21:39 these, again, API keys within your\n21:41 system. Giving Hermes the ability to\n21:43 connect to anything. Anything that has\n21:45 an API key, you can connect to Hermes in\n21:47 exactly the same way. And what's cool\n21:49 about this is Granola also has an OAuth.\n21:51 So, if you just give it the name of the\n21:53 Granola MCP like this, it will actually\n21:56 ask you to authenticate and sign in. And\n21:57 look at this, guys. Hermes Agent would\n21:59 like to access your account. And the\n22:01 Hermes Agent actually opened up this MCP\n22:04 MCP server for me. So, I come down and\n22:06 allow access. And you can see\n22:07 authorization successful. You can now\n22:09 return to the CLI. And all I did, guys,\n22:11 honestly, I just came down. I copied\n22:12 this page. I just sent the link and\n22:14 shared it straight to Hermes. And as you\n22:15 can see, it's come down here and it's\n22:17 confirmed that it's Farm Blast Meeting,\n22:18 which is a community call I had, and\n22:20 exactly one of the questions that we\n22:22 asked. And that's And it's such a\n22:23 wonderful example of how you can connect\n22:25 your Hermes agent to your entire world.\n22:27 So, I could be anywhere and ask it any\n22:29 question I like. My concept 11 is the\n22:31 fact that Hermes lives on your computer.\n22:33 If you've set it up this way. Which\n22:35 means that not only can it do things\n22:36 like create files on your desktop or\n22:39 organize things on your desktop, it can\n22:41 also take actions on your computer. So,\n22:44 for example, it can do what? It can do\n22:45 browser vision, bash, fuse. What the\n22:47 hell does that mean? It has native\n22:49 Chrome DevTools, visual screen reading,\n22:51 real cursor movements, not a screenshot\n22:53 bar, an actual agency. So, if you're\n22:55 running to the gym and you're like,\n22:57 \"Damn, I completely forgot this thing.\"\n22:58 Basically, Hermes Agent can go ahead and\n23:00 do that. Now, to be fair, this isn't a\n23:02 feature I tend to use a lot, but again,\n23:04 if you just think about the fact that\n23:06 the agent lives on a computer, it can\n23:07 take actions and do various different\n23:09 things for you once you enable it with\n23:11 the tools. Concept 12 is around about\n23:12 using MCPs. MCPs stand for model\n23:16 contacts protocol. So, think of it like\n23:18 this, okay? You have an an API is like\n23:21 the wiring, and the MCP is like the\n23:23 instruction manual around that wiring.\n23:24 So, instead of having like a like, you\n23:26 know, a remote, for example, with a lot\n23:27 of exposed wires, MCPs are basically\n23:30 like instruction packages. It's like a\n23:32 universal remote control that lets you\n23:34 connect Hermes to effectively anything.\n23:37 We just used the Granola MCP to connect\n23:40 to basically my meetings. And we can do\n23:42 the same thing, for example, with Zapier\n23:44 MCP. So, Zapier MCP lets you spin up\n23:46 these servers like so, new MCP server,\n23:49 and I can go ahead and I can connect to\n23:51 anything that I wanted, right? If I come\n23:52 down here, for example, let's say that I\n23:54 want to come down and do other, I come\n23:56 over here and I connect to things that,\n23:58 you know, through this you can't connect\n23:59 to other applications. I can add tools.\n24:01 So, for example, I wanted to give it the\n24:02 ability to read emails, right? Well, how\n24:04 would I do that? Well, we can just\n24:05 basically select all the stuff that you\n24:07 want it to do. The one thing I wouldn't\n24:08 give it access to is the ability to send\n24:10 emails just yet. I would just hold off\n24:12 on that and its ability to reply. And\n24:14 when that's done, I can add these 10\n24:16 tools. I can add in things like Google\n24:18 Drive, I can add in things like whatever\n24:19 I want to, Zapier. It's really cool.\n24:21 Then, all I do, I've got skills over\n24:23 here as a second tab. But, what we're\n24:25 really going to want to focus on here is\n24:26 the connection. And basically, what we\n24:27 can do now is generate a token here. And\n24:30 I've got an MCP server to connect to\n24:32 Zapier, which is awesome. Now, what I\n24:33 can do is come back over to Hermes, and\n24:35 I can say, \"Hey there, I'd like to\n24:36 connect to the Zapier MCP server. I've\n24:39 got my tokens. Could you give me a\n24:40 command to run through the terminal so I\n24:42 can add these to your environmental\n24:44 variables?\" Again, you can paste it, but\n24:46 I I'd often recommend that you do this\n24:48 in the terminal so it doesn't get\n24:49 recorded in your chat logs. And once\n24:51 you've done that, you can say, \"Awesome,\n24:52 I've done. Could you tell me the title\n24:54 of the last email I received?\" Again,\n24:56 then it will use an MCP to go ahead and\n24:58 do it. Now, why would you create an MCP\n25:00 versus just using an API key? It is way\n25:03 more what we call token efficient\n25:05 because all the checking it needs to do\n25:07 builds up the cost. An MCP basically\n25:10 explains exactly how it works so AIs can\n25:13 basically just look at the MCP manual\n25:16 and they'll know exactly all the stuff\n25:17 they can do, all the buttons they can\n25:18 press in that individual application.\n25:20 So, it knows how to draft emails, create\n25:23 emails. It has full understanding\n25:25 because it standardizes the use of that\n25:28 incredible tool for the agent. And just\n25:30 like that, we've got all the information\n25:32 that we need from this MCP. Concept 13\n25:34 is understanding muscle memory for the\n25:36 agent. The idea here is that the longer\n25:38 we use Hermes agent, the better it gets.\n25:40 What I mean by that is consistently give\n25:43 it feedback as you go so it can iterate\n25:46 and improve as you're doing it. It has a\n25:47 series of Hermes skills it can pick up\n25:50 the more that you use it. It has various\n25:52 other things it can do and you can even\n25:53 if you want to build out different\n25:55 skills quite easily. So, for example,\n25:57 you can go to this website here, which\n25:58 is agentskills.ai. This gives you a good\n26:00 overview of what skills are if you want\n26:02 to go deeper on that. One of the things\n26:03 that I did actually guys is I built out\n26:05 something called the Pantheon, which\n26:06 effectively enables me, say for example,\n26:08 I want to add a persona, right? I can\n26:10 choose let's go with the alchemist for\n26:11 instance. I can explain its job. I can\n26:14 explain its description, give it a\n26:15 system prompt and then actually choose\n26:17 the model that I want to do it with\n26:19 because again, if it's like very\n26:20 intellectualized, we don't want Albert\n26:22 Einstein, you know, washing our car for\n26:24 example. You know, we want him working\n26:26 on the most impactful tasks. And so, you\n26:28 can build out skills like this. Look, I\n26:30 have a deep reasoning task I use for\n26:31 Claude, but you know, for autopilot\n26:33 jobs, I might use a lower cost model.\n26:36 So, you can build out these skills which\n26:38 give it the ability to do something\n26:39 different. So, for example, I could say\n26:41 something like, \"Hey there dude, I'd\n26:42 like to do some research for me, some\n26:44 very deep research on the best country\n26:46 in the world to live. Um you know, spin\n26:47 up a sub agent. Why don't you go ahead\n26:49 and use Deep Seek V4 and ChatGPT 2\n26:53 critically analyze it in a nice little\n26:54 loop with Opus 4.7 reviewing everything.\n26:58 Um, just let's go ahead, use a cheap\n26:59 powerful model to go ahead and do loads\n27:01 of deep research, take some time with\n27:02 it, and then come back and share the\n27:04 results based on everything you know\n27:05 about me. Now, what's cool about this is\n27:07 it can delegate to different models to\n27:09 do different things for me, and it can\n27:11 use that as a skill. Now, it will use\n27:13 here for me the deep research skill that\n27:15 I've already created, which is fantastic\n27:17 and it just enables you to get so much\n27:19 more out of Hermes. Now, if anytime you\n27:21 want to stop it from doing a task, you\n27:22 can do {forward slash} stop, and that\n27:24 will actually basically stop Hermes from\n27:25 doing the thing. The concept 14 is about\n27:27 six keys to make it so unbelievably\n27:30 powerful. So, here we've got we've got\n27:31 Q. Now, what's interesting thing about Q\n27:33 here, okay? Now, Q is this, which means\n27:36 that let's say I ask it a question like,\n27:38 \"Hey dude, find out for me the best\n27:40 country for octogenarians. Which country\n27:43 world has the highest number of\n27:44 octogenarians?\" Okay? Really cool. I can\n27:46 send it off. Then let's say I want to do\n27:48 a follow-up task, I can do {forward\n27:50 slash} Q. And what this will do is send\n27:53 a prompt on the next turn so it doesn't\n27:54 interrupt. I can also do {forward slash}\n27:57 background, okay? And what {forward\n27:58 slash} background lets me do is\n28:00 effectively ask it let it do different\n28:02 tasks in the background. So, for\n28:03 example, I could come down and do\n28:04 background, and then I might come back\n28:06 and say something like, \"Hey, find out\n28:08 which country in the world has the\n28:10 healthiest 35-year-olds or\n28:12 32-year-olds?\" Okay? Now, this will run\n28:15 in the background. Cuz normally when you\n28:17 use Hermes agent, basically what's\n28:18 happening is it's it's interrupting. So,\n28:21 this enables us in one chat window to\n28:23 have met multiple streams of thoughts\n28:26 running concurrently by using those two\n28:28 tasks. And as you can see, background\n28:30 task has gone ahead and actually\n28:31 physically got started. You got Kanban,\n28:33 which is for a a task board\n28:34 functionality you can have. Reset will\n28:36 clear it all up. Compress will compress\n28:39 the memory. So, important thing to bear\n28:41 in mind on the memory is that there is a\n28:42 certain size of the window. So, if you\n28:45 have, say for example, 10,000 words, and\n28:47 you ask it a question, every time you\n28:49 ask that chat agent a question, it is\n28:51 literally using all 10,000 words plus\n28:54 your question. And the longer it gets,\n28:56 basically, the lower its performance\n28:58 gets. So, what I do is one session, one\n29:00 window done, you want to reset. So,\n29:02 compressing basically means, \"Dude,\n29:03 could you just do me a favor and just\n29:05 summarize everything we talked about?\"\n29:07 So, it goes from this to like this,\n29:09 basically, and then you can continue the\n29:11 conversation. And as I showed you\n29:13 earlier, we can switch the models super\n29:14 easily. And what we can go down here and\n29:16 do {forward slash} model, and then we\n29:18 can basically Now we're talking to Opus,\n29:20 we're talking to ChatGPT, we can talk to\n29:22 anyone that we physically like. And I\n29:23 put a lot of the interesting ones down\n29:25 here that I thought were really quite\n29:26 interesting that you may find yourself\n29:27 using, you know, basically depending\n29:29 upon what situation you're in the\n29:30 moment. Now, concept 15 is what we call,\n29:33 basically, safety and security for the\n29:34 agent. These are house rules. Now, these\n29:36 agents have access to many different\n29:38 things, and because of that, I think it\n29:39 was Peter Parker told us this, right?\n29:41 With great power comes great\n29:42 responsibility. Might have been Ben's\n29:44 uncle.\n29:45 >> Remember, with great power\n29:48 comes great responsibility.\n29:50 >> But the point here is we basically don't\n29:52 want to give it the keys to the kingdom\n29:54 because that's responsible years, which\n29:56 is why I don't let it have the ability\n29:58 to send the emails to you. A lot of my\n29:59 buddies have done that, they've had\n30:00 their hands burned in the process, and\n30:02 it will get to the point where it will\n30:03 have less errors than humans. But at the\n30:05 moment, we want to follow what's called\n30:07 the principle of least access. And what\n30:09 that effectively means is Herme Herme's\n30:11 and any agent or system or anything\n30:12 you're building ever. I do this with all\n30:14 my clients. You only ever give it the\n30:16 minimal amount of access that it\n30:18 actually needs. The very, very minimal\n30:21 thing. So, for example, there's certain\n30:22 things that we don't want it to go ahead\n30:23 and do. API keys, never, ever, ever post\n30:27 into any chat like a Herme's especially\n30:29 specifically if it's keeping it. It's\n30:31 not that Herme's is bad person, it's\n30:33 just that if you're backing up your\n30:34 Herme's on a daily basis, we just don't\n30:36 want these keys. It's just one less\n30:38 thing for you to think about. So, if you\n30:40 can actually think about using it\n30:41 responsibly like that and follow this\n30:43 principle of this access, you will\n30:45 generally across all of your AI systems\n30:47 be extremely safe. Now, next thing to\n30:49 think about here in concept 16 is called\n30:51 the North Star and this is what's called\n30:53 the goal function. So, think about like\n30:55 this, okay? Rather than going question\n30:57 answer question answer, the idea of the\n30:59 goal functionality within Hermes is that\n31:01 you give it a North Star for a\n31:02 short-term project. So, I could say,\n31:04 \"Hey, I want to create a YouTube\n31:06 thumbnail that is pretty freaking epic.\"\n31:09 So, for example, I I come over to\n31:10 Hermes, right? And I come down here and\n31:12 why don't I just go ahead and clear this\n31:14 entire session. And so, for example, you\n31:15 come to the Hermes agent and you can\n31:16 give it a prompt like this, which is\n31:18 goal. I would like to decide whether I\n31:19 should take an opportunity A or B.\n31:21 Should I do a big consulting role at\n31:23 Anthropic or do I become head of\n31:24 innovation at OpenAI? I would like it to\n31:26 help me out. Help me make the best\n31:28 decision. And look at this. Now, it's\n31:29 come and said, \"Look, I'm going to\n31:30 assign a 20 goal budget as well a 20\n31:33 turn budget to this.\" And effectively,\n31:35 what will happen now is our entire\n31:36 conversation exists to hit and crush\n31:40 that specific goal. So, it will ask me\n31:41 questions about what's about, what's\n31:43 going on, and it will get there on a\n31:45 mid-term basis. Now, what's really cool\n31:47 here is Hermes will not stop until it\n31:49 hits that decision. But, it does run\n31:51 into one limitation and that's the fact\n31:53 that not all goals we do in life can be\n31:55 solved in 20 turns and sometimes we've\n31:57 got to do things. This is why I built\n31:59 this idea of super goals and I'll put a\n32:01 link on screen if you want to look into\n32:02 this. The idea of super goals is that\n32:04 you explain the goal you want to, use\n32:06 the super goal skill, and effectively,\n32:08 what happens here is Hermes comes and\n32:10 asks you loads of questions and the\n32:12 output of that is like say four to 10\n32:15 actions. Some that you do, some that\n32:17 Hermes does. For example, on this one,\n32:19 which is to launch a Hermes course and\n32:21 get 500 sign ups, one is shape the\n32:23 course promise, build an opt-in engine,\n32:24 write course sequence. Then, it's my\n32:26 turn. I have to go ahead and record the\n32:28 video promo and it lets me mark this\n32:31 task as complete when it's done and then\n32:33 my progress bar goes up. But, if I were\n32:34 to remove this and start from scratch,\n32:36 you'll see right now, look, every\n32:37 hearing is a great goal. I can copy this\n32:39 prompt. I can shoot straight back over\n32:41 to Hermes. What I can do is {slash} and\n32:43 I just come down here and click on stop.\n32:45 As if for instance, say gold pause is\n32:47 what we want to do. And the command I\n32:48 believe is goal clear. This will end the\n32:50 goal. And what I can do is come down,\n32:52 give it this information here, and in\n32:53 effect from now what this does will\n32:55 allow me to have a medium-term goal. And\n32:57 I'll say, \"Hey Jack, what is the goal\n32:58 that you want to go ahead and set?\" And\n32:59 so say, \"Hey dude, I would like to\n33:01 launch a paid course and get 1,000 uh\n33:04 email sign-ups to my new Hermes\n33:06 masterclass.\" Or whatever the thing is\n33:08 that you want to sell or or help people\n33:10 with. And then it starts to ask\n33:11 questions. I'll say, \"Hey, in 4 weeks,\n33:13 okay?\" And then it asks you questions\n33:15 dynamically. And I tested this so much,\n33:17 which is why I built all these different\n33:18 specific rules into it. And it's all\n33:20 connected to the dashboard. So when this\n33:21 is done, it will literally load up in\n33:23 the dashboard dynamically. And again,\n33:24 ask you different questions to get\n33:26 context. Cuz without context, how can it\n33:28 know that it's helping you out? Then all\n33:30 you ever do is you land on your mission\n33:31 control, you command it in Hermes, or if\n33:33 you wanted to, you can come down and use\n33:35 Claude. And it's the exact same thing,\n33:37 so you can pick up whenever you want to.\n33:39 Honestly, these mid-term goals, these\n33:40 chief working loops, are so freaking\n33:42 powerful. You can plan your life, hit\n33:44 these mid-term goals using Hermes agent.\n33:46 But concept 17 is about using Hermes\n33:48 agent as a team. So the idea is that we\n33:50 can tag in different models to various\n33:52 different things. You can spin up what\n33:54 is called sub agents. So for example, if\n33:56 I come back over now to Hermes agent,\n33:58 what I might do is just say, \"Hey, I'm\n34:00 just going to pause that for a second.\n34:01 What I would like to do is go ahead and\n34:03 research for me the best AI companies to\n34:05 work for on the planet. I'd like you to\n34:07 spin up two sub agents. One is going to\n34:09 go ahead and look at the ones\n34:11 in America, and the other one is going\n34:12 to look at alternative jobs maybe\n34:14 internationally, okay?\" And what Hermes\n34:16 can do is spin up sub agents. And what\n34:20 sub agents are is think about like this.\n34:21 You're at Hermes, and then underneath\n34:22 Hermes you have two other agents, each\n34:25 with their own fresh contacts. And it\n34:27 delegates these tasks, these two agents\n34:29 go ahead, they do all the work, and they\n34:31 report back. Now, this is happening in\n34:34 parallel, not sequentially, which means\n34:36 that I could have four, five, six\n34:38 different agents all researching things\n34:40 like a team doing work individually and\n34:42 at the end of the day or the end of the\n34:43 turn, everyone hands in their paper,\n34:45 their research, and instead of waiting,\n34:47 you know, six hours, it's done in one\n34:48 hour instead. This is called\n34:50 parallelization, parallel agents, and\n34:52 gets you significantly further faster\n34:55 using Hermes. I use this strategy all\n34:57 the time to get the most out of the\n34:58 agents. For example, here, Hermes\n35:00 co-founder Teknum, 12 parallel Hermes\n35:03 agent enters. I hope I pronounce his\n35:04 name correctly there. Every day to build\n35:06 Hermes itself, monitoring issues, dog\n35:08 food in the Kanban good board, parallel\n35:10 sub-agents are the power user move. So,\n35:12 even the guy that built Hermes agent\n35:14 used parallel agents to do it. So,\n35:16 again, multiple agents working together\n35:18 in tandem to one objective. Or it could\n35:20 be individual objectives, right? One\n35:21 could be doing research, one could be\n35:22 writing, one designer, one scheduler,\n35:24 and they all put these beautiful folders\n35:26 inside your desk. The concept 18 is so\n35:29 powerful. This was called the heartbeat\n35:32 or the cron job. Cron job sounds so\n35:34 freaking complicated. One of the coolest\n35:35 things you can do is this idea that it\n35:37 keeps working whilst you sleep, the 24/7\n35:39 AAA agent, okay? A heartbeat effectively\n35:42 is something that you set up to happen\n35:44 every single day. It's essentially\n35:45 called a zombie as well, a zombie\n35:47 process, if you like. And effectively,\n35:49 what it will do is you give it something\n35:51 to do. So, effectively, a heartbeat is\n35:53 silent until something dies. So, a\n35:55 sub-agent pings every few seconds,\n35:57 zombie detection, fresh agent reclaimed\n35:59 to task. Per task retries with\n36:01 hallucination recovery. What does that\n36:03 mean? Let me show you what a cron is\n36:05 first, and then I think it'll make a\n36:06 little bit more sense. So, a cron agent\n36:08 here, if I come down here and just say,\n36:09 \"Hey, dude, I want you to do two things\n36:11 for me. I want you to send me a reminder\n36:12 in 30 seconds that the sky is blue,\n36:15 okay?\" This is a scheduled task. This is\n36:18 basically in a certain amount of time,\n36:20 it is going to come and give me an\n36:21 update. Now, for example, I have a daily\n36:24 morning briefing, okay? And I say, \"I\n36:26 want you to go through my um, email\n36:27 list, I want you to go through my\n36:28 calendar, I want you to go through\n36:30 everything I'm doing that day, use all\n36:32 the information that you know about me,\n36:34 and give me, you know, a daily morning\n36:35 brief of, say, I don't know, five things\n36:38 that are relevant. And it can build that\n36:39 for you. So, every day you get that on\n36:42 repeat, basically. So, you can schedule\n36:43 these wonderful things inside Hermie's\n36:45 Agent, get to come to you at 8:00 a.m.\n36:48 every single morning. But, the one that\n36:49 you're going to find yourself using most\n36:50 of the time are these scheduled tasks\n36:52 that make it feel like it's alive\n36:54 because proactively coming to you and\n36:56 asking questions. You could say, \"Hey,\n36:58 uh, periodic times randomly come and ask\n37:00 me questions to check in on me, do a lot\n37:02 of other things.\" Then, we go to concept\n37:03 number 19, which is essentially the\n37:05 dirty secret not a lot of people want to\n37:06 talk about, which is, can you believe,\n37:08 73% of every quest is fixed overhead.\n37:11 So, what we're talking about here is\n37:13 budget, okay? Now, what I mean by that\n37:15 is that, like, if you are using API\n37:17 keys, you can very easily burn through a\n37:20 lot of money, which is why we want to be\n37:21 very specific about the specific models\n37:23 we're using. Now, tokens, you know, um,\n37:25 a parking meter takes maybe dollar\n37:27 coins, um,\n37:28 AI models take tokens. Roughly speaking,\n37:31 it a token, I guess a 10 tokens is about\n37:34 seven words. It's about seven seven 70\n37:36 75%. That's effectively what the\n37:38 translation looks like. So, just make\n37:40 sure that you often clear the session\n37:42 and that you're using the right model\n37:44 for the right job when you're using\n37:45 Hermie's to keep your token cost usage\n37:48 all the way down. And there's loads of\n37:49 mini strategies, like, don't overload it\n37:51 with a bazillion skills that you don't\n37:53 need, don't make it system prompts too\n37:55 long, keep it nice and clean, and one\n37:58 conversation, one goal, then it's always\n38:00 best practice to clear, and refresh, and\n38:02 start again. And just point it out, some\n38:03 guy actually spent 4 million tokens in 2\n38:06 hours of light usage, 21,000 tokens,\n38:09 just asking about the weather because he\n38:11 made a mistake. So, it just gives you a\n38:12 good a good idea of how you can reduce\n38:14 those down. Now, the 20th concept here\n38:17 is this idea that it isn't just a\n38:19 chatbot, Hermie's Agent can be part of\n38:21 an operating system. Well, we have\n38:23 various different personas in the\n38:25 Pantheon. We can have a memory vault\n38:27 that extends its memory capabilities to\n38:30 a significant level, basically enabling\n38:33 you to call upon any email or meeting\n38:35 that you've ever had, any knowledge\n38:37 source from any YouTuber you've ever\n38:38 seen,\n38:39 connected to notebook 11, a\n38:41 gargantillion different things. We can\n38:44 have, for example, so many different\n38:45 tools and effectively when we create an\n38:47 operating system, it just gives it a\n38:49 place to manage everything. I call this\n38:51 a kind of everything of AI. So, if you\n38:52 look at this Claude code operating\n38:54 system, we have our AI spendage, how\n38:56 much money we're making from our skills,\n38:57 we can see our usage in real time,\n39:00 dreaming improvements. Like I said, they\n39:01 will come to you proactively, mission\n39:03 control, our skills and memory systems.\n39:05 Now, in the Hermes operating system\n39:08 here, I can see at a glance all of the\n39:09 connections it has available to it. So,\n39:11 I know exactly what it's connected to. I\n39:13 know what model I'm using, I know my\n39:14 memory, I know my usage. I can chat to\n39:17 it directly. I have this ability to\n39:19 manage goals and plans with Hermes\n39:21 agent. So, whatever I'm working on, I\n39:22 can work with Hermes agent in tandem on\n39:24 that. I have this Pantheon, so I can\n39:26 create these beautiful I don't know I\n39:27 personally like to visually see my\n39:30 graphics and what I'm dealing with. So,\n39:31 I can like, well, actually now I really\n39:32 want to kind of upgrade that like I'm\n39:34 missing this or I'm missing that. I can\n39:35 grab different things and build various\n39:37 different things. I just think it's a\n39:38 nicer user interface to be able to go\n39:40 ahead and build things like that. Again,\n39:42 and what we can do here crucially and I\n39:43 highly recommend you do is connect your\n39:45 Hermes to GitHub. Now, GitHub is a\n39:48 basically code for place where we store\n39:51 stuff. Okay? It's like the world's\n39:52 repository. It's basically you can store\n39:54 information and files. And what you can\n39:56 do with Hermes agent, guys, is\n39:57 essentially have your entire Hermes\n39:59 agent backed up on GitHub every single\n40:02 day. Mine does that. And what that means\n40:04 is if you ever change your computer or\n40:05 God forbid you lose it or something, you\n40:07 have that backed up somewhere, right? We\n40:09 can connect to GitHub. We've got all the\n40:10 profile information here. We can connect\n40:12 to our Obsidian brain, which is this\n40:14 beautiful memory system here that\n40:16 basically just lets us see everything\n40:18 we're doing with different memory\n40:19 systems. We've got there's my Claude and\n40:21 it just enables you to get a level of\n40:22 control. And crucially it comes on to\n40:25 the 21st concept which is the idea of\n40:28 one brand. The idea that we are going to\n40:30 connect the Hermes agent, the business\n40:33 brand, with Claude code. We talked about\n40:35 the differences but here's the\n40:36 difference guys. If we can actually\n40:37 connect them together, it unlocks\n40:39 incredible capabilities. Hermes knows\n40:41 your projects, your customers, your\n40:43 decisions. It lives everywhere.\n40:45 everywhere. Maybe it's like cotton who\n40:46 knows what it is. And then you got\n40:47 Claude code is this precision tool that\n40:48 we use for building. And actually what\n40:50 happens is when we can physically\n40:52 connect these ones together, we have\n40:54 this shared memory and understanding\n40:56 where we're not operating in isolation.\n40:59 It's like you say one thing to Hermes\n41:01 agent, another to Claude code. How do\n41:03 you know they connect together in a\n41:04 meaningful way unless they have access\n41:07 to basically what they're doing at\n41:08 different times. And so now we've\n41:09 covered these 21 core concepts in plain\n41:13 English. The next thing we need to do is\n41:14 leverage all of these and get one of\n41:16 these operating systems ourselves which\n41:18 you can learn about in this video right\n41:20 here."
}