Everything New in Hermes Agent (Full Setup Guide)

0:00This video will teach you 90% of Hermes agent and cover everything that you need to know. We'll go through all of the important features. I'll show you how to set it up completely from scratch, and I'll go over all of the new updates, because since my last videos. A lot has changed and this has become even more powerful.

0:16So with that said, let's dive in. All right. So by the end of this video, this what we'll have now, we'll have a server that's running Hermes agent 24 over seven. I'll show you how to install the desktop application and how to chat with it from your phone. I'll show you how to set up scheduled jobs that run automatically. And I'll show you how to set up

0:32multiple different agents and bot mode, which is a new feature. Now, like I mentioned, there are actually a lot of new features in the most recent version of Hermes, which is why I'm making a new video on it. For example, the desktop app, which you probably haven't seen yet, bought mode, which is really cool

0:47and allows you to set up different Hermes profiles connectors. So things that you're used to in like cloud code built directly into Hermes agent, local models, meaning you can directly run models. It's pretty easy to do that plug in catalog. So there's a new plugins that have been added. And then of course we have the same old stuff

1:03like schedule jobs, subbasins, password managers, all that kind of stuff. So I'm going to go through all of that. You're going to see exactly how it works. And I'm going to try to cover this as comprehensive as possible and get through 90% and pretty much all the important features. All right. So let's begin by just discussing what Hermes agent actually is. Now, once we understand that, I will give you a full, in-depth tutorial

1:24of how to set it up and deploy it at all the time. Stamps for everything I go through will be available from the link in the description. Okay, so Hermes agent is effectively an AI agent that runs on its own computer. Now. It can take its own actions, which means it can browse the web,

1:39it can call APIs and it can do things for you. Now, importantly, it also remembers you. So it will keep notes about you. It will save memory and it will actually improve over time. And that's kind of the last point here. It learns which is a big differentiator between a lot of other agents. It will figure out what tasks you're completing, how you're completing them, and it will improve over time by learning everything

2:00that you're asking it to and just progressively getting better. Hermes agent is really good when you're actually going to be chatting with a day to day and using it as one of your primary agent sources. It's not something that I would typically just let run in the background 24 over Now, while you can do that, you get most of the value out of this

2:16when you actually chat with it, teach it, it learns from you. You get the idea. Now it just created a quick comparison chart here so you can see the difference between something like Claude Code ChatGPT whatever and Hermes Agent. Basically, the summary here is that this is running 24 over seven all of the time. It can do tasks, automations, and is really good for things

2:35that are going to be running recurrently. You can also connect to it from your phone, discord, pretty much any channel that you want. And importantly, it writes and improves its own skills and gets better over time. Now both Claude Code and Codex or Claude whatever will somewhat do this, but Hermes works a lot better in my opinion, and is a lot more configurable.

2:54And importantly, you can swap the models in and out. So if all of a sudden GPT is the best model, you just switch to that. If Claude is the best model, you go to that. If you want to use the local model, you do that and you're not tied into any company's pricing. You own your own agent on your own deployment.

3:08So you have full control, which I think is the biggest selling point of it. So I want to go through the six part of Hermes agent, just so you can briefly understand what they are and how they work. And then we're going to get into the setup. Now, basically, Hermes agent is a harness. What that means is that it's really a set of tools and kind of a framework

3:26that lives around an AI model or an LM. Now, with Hermes agent, like I discussed, you can pick any model that you want. You can use open router, you can use a cloud model, you can use a local model, you can use deep seq. And that's one of the advantages of now you can plug these models in

3:41and switch between them and even use them for different tasks. Next we have tools. Now these are things that Hermes itself will use. It comes with a bunch of prebuilt tools to do things like browse the web, right skills, modify files, search for them, etc. but you can install your own tools as well.

3:57And to be honest, the tools that you add will make a massive difference. And I have all kinds of videos on this channel discussing the best tools to use for AI agents. Next is memory. Hermes agent has its own memory system. This allows it to automatically save information about you, remember sessions, transcripts, and important context and search through that memory

4:17so it doesn't need to load all of it at the same time. Next, we have skills. This is probably the biggest differentiator of Hermes. Agent Hermes will automatically create, modify, and update skills for you, which are reusable markdown files that instruct the model how to perform tasks. So if you tell the model, hey, go do this accounting task for me, it will usually automatically save a skill related to that task, and the next time you ask it to do that,

4:43it will simply pull from its memory, read the skill, and know how to do that. Now, if you refine it, or you update it, or you give it an improvement or some feedback, it will adjust the skill automatically. And that's how it kind of learns and gets better with you. Next we have gateway.

4:58Now the gateway refers to how you communicate with Hermes. There are a ton of different gateways you can connect. One of the most popular ones is telegram, but you can also connect it to discord. You can connect it to WhatsApp, you can connect it to pretty much any app that you can think of, and I'll show you how to do that.

5:14And then lastly we have cron. If you hear this term, this just means something that's running on a schedule. So a task that's running every day or every hour or every Monday or every weekday at 5 p.m., you get the idea. So these are kind of the core concepts. We'll go through all of them one by one. But I just wanted you to get a kind of piece of the picture here of how this fits together and how Hermes

5:34operates. Okay, so how does Hermes learn? This is something that I kind of already discussed, so I'll go through it quickly. But basically when you ask Hermes to do something or you give it a task, it will figure out by itself how to perform that. It will then save a skill that outlines what it did, and then the next time it needs to go through this process, it will be faster because it doesn't

5:54need to figure out that process again. Now, if you refine it or you tell it to fix something or the way it did, it is incorrect. It will update that skill and it uses a process or something internally called a skill curator, which effectively does this learning. You also can manually tell it to learn something based on a task

6:12that you've given it. And again, this is one of the main differentiators. A lot of other harnesses are copying this now, but this was one of the first ones to come out with this idea, which is why it got a lot of adoption. Okay, so let's keep going here. And now talk about how to run and deploy Hermes.

6:27Now I run Hermes personally on a virtual private server. And this is what I will show you how to do in this video, because it's the recommended approach. That said, you can run this locally on your own computer if you have your own dedicated hardware device, but typically that's not going to be cost effective or the most secure option. So let me quickly explain why I run this on a virtual private server

6:46and what that is. So a VPN or a virtual private server is its own dedicated computer that's running typically inside of a data center. Now, this is much more secure than something like your own local computer. It has a firewall, for example. It has its own dedicated network. And importantly, you can turn it on and off, delete it, and stop paying for it

7:04pretty much whenever you want. Now, if you were to buy your own 600 or $700 device to run something like Hermes locally, that's fine. But of course you're committed to that. You need to do all of the security setup, and if something goes wrong, it's a lot more difficult to just nuke that device then you can with something like a VPN. So from the security angle, in my opinion, this is more secure and it's set up

7:24by default to be more secure than your own local computer. Importantly, you can also rent it for just a few dollars per month, so it's a much more affordable option than buying your own thing or buying your own hardware device. And this will run 24 over seven. So whether you're on your phone or a different computer or whatever, you don't need to worry about the power

7:41or the network or you keeping this device online. It will just be online. It will not go offline unless you turn it off manually. It also has automated backups, all this kind of stuff. And again, I talked about this in many videos, but it is the recommended approach that I personally use for all of my agents,

7:56specifically things like Hermes agents where you always want them to be online. So with that in mind, I will show you the provider that I personally use and that I recommend now that is hosting her. And I do have a long term partnership with them. They sponsor a lot of my videos. So that allows me to give you guys an additional 10% off. One of the reasons I like to use this is they have a one click deploy option,

8:16which means you can simply just go to this web page. I'll leave it down below. Choose a plan you want. I typically recommend going with the KVM two plan, and then it will just automatically deploy and set up Hermes agent for use that you don't need to do any of the steps yourself. Now of course, use whatever provider you want. Ultimately, I don't really care, but if you guys want one of the best options,

8:35then go with hosting or from the link below. You can use the discount code tech with Tim, which will give you an additional 10% off, and then choose your plan duration here. If you want the best price, then of course go with a longer plan. From here you can uncheck this Nexus AI thing. I typically don't use that, but if you want to go

8:50ahead, choose your server location and then go to continue and deploy. It will give you a pop up after this with some information. Don't worry, you don't need to save any of that data. I will show you how to get to it later. Proceed and then it will automatically deploy and set up Hermes for you. Now when you go to deploy this,

9:05you should see a pop up that looks something like this. If you don't, don't worry, but it's probably going to be something that kind of looks like this. Dashboard. Basically, you can just press next from here. In my case, I'm just deploying this on a I already own because I already have a bunch of them. And then simply press deploy and it will take a few minutes,

9:21set everything up for you, and then I'll show you what to do afterwards. If you don't want to run this on a and you are going to put it on your own computer, I will leave a link to the documentation down below on how to do this. You can install the new desktop application which will set up a lot of stuff for you, but keep in mind that is running locally

9:37and will be turned off if you turn off your computer. And you also can obviously use a different provider and then go through the installation setups here with something like the curl command, or then you'll run through the full flow in our case hosting, or will just set this up automatically for us. And then I will show you how to access it, go into the config.

9:53And that's what we're going to do next. So to find the password you're going to look for this Hermes container again from Docker Manager. And you can press on manage from manage. It should show you the details of this application where you can then see environment. And from environment you should see a field

10:09that says admin password as well as the admin username. So what you can do is just view this. I'll delete the container after. So don't worry if I'm leaking the password. And what I'm going to do is just paste it here and then sign it. Now this will verify that Hermes is working, and what you'll be able to do is be brought to the Hermes

10:25dashboard here, where you can start chatting with Hermes and using it. That said, there is still some configuration that we're going to want to do here in order to use the agent and to add different models, and we probably want to access it from the command line as well. So I'm going to show you how we can do that, because sometimes

10:41you won't just be able to access it from the web dashboard like this. Okay. So let me go back to my presentation here, and I want to quickly explain how Hermes runs on this particular virtual private server, just so you have an understanding. And again, if you install it yourself, then pretty much everything will be the exact same, minus

10:57the Docker container stuff that I'm going to go over here. Okay. So basically we have our virtual private server right. In this case we're using Host Stinger. You can use whatever you want. Now within this virtual private server we have a piece of software called Docker. Now Docker allows us to create these managed containers

11:14which are kind of like isolated environments and applications. And in this case, our Hermes agent is running in what's called a Docker container. Now, the reason you need to know that is in order to actually access the Hermes software on the virtual private server, you'll need to be inside of that Docker container. So that's why you saw we went to Docker Manager.

11:34There's kind of an environment for this Docker manager. And we also have the ability to go into the terminal and stuff, which we're going to look at in a second. But just understand that on this it's running inside of doc. Now in order to manage Hermes and to use it, there's really three main ways I'm going to show all of them to you. Now. The first is the browser, like we just saw over here where we kind of signed

11:55in, and we can start chatting with it and using it directly from here. The second is the terminal, which means we can actually view Hermes from our terminal environment, which in this case will require we do something called SSH, which is essentially a secure connection to the server. Again, I'll go over that in a minute. So don't worry. And then another way that we can manage Hermes is we can actually install

12:16cloud code onto the server and simply ask cloud code to manage it for us. Now, the reason I'm going through these different options here is that sometimes there'll be something that gets messed up with Hermes, or you want to do something and you're not sure how to do it. If that's the case, rather than struggling with it. What I typically recommend is that you install something like codex

12:37CLI or clone code CLI directly where your Hermes is installed. So either on your local computer or on something like a virtual private server. That then means that if there's a mistake or Hermes goes down or you can't figure out how to fix it, you can literally just tell Claude code, hey, fix this thing.

12:54Hey, Hermes isn't running. Can you fix the correct command? Can you set it up for me? And it can actually manage it and run commands on the server to set it up for you? So again, right now we've ran it from the browser. We're in Hermes. I'll look at this later and show you how it works. We're also going to show you

13:09how to do it from the desktop application, which is super cool. But then you can also run it from the terminal. And then when you're in the terminal, if you have a mistake, you can run it with CloudKit. So first let's see how we access it from the terminal. Now again, if you've deployed with the VPN, what you can do is

13:25go to Docker Manager applications and find your Hermes container. From here you can simply press on terminal. When you do that, it's going to open up an SSH session and allow you to directly access Hermes from the command line. So you can see here we're actually inside of what's called the Hermes Docker container.

13:45And what I can do is I can run the Hermes command. Now, if I run the Hermes command here on this virtual private server, we can just press. No, thank you for this. And it's going to open up the Hermes terminal user interface where I could then, you know, chat with it directly from here.

14:00Okay. So that's how you access it from the command line. Now I'm just going to exit out of that by hitting control C on my keyboard. Now from the command line we also have the ability to say Hermes and then set up or to run any other commands, which we'll look at later in this video to configure Hermes. Because remember I told you that with Hermes agent, we can connect different models,

14:22we can set up different tools or MCP servers or plugins or configurations. A lot of that you can do from the visual app, right. So from like this web UI here, but some of it does require that you're in the command line. So that's why I'm showing you how to get into the terminal. So I know it's a little bit confusing, but I just want you to understand okay,

14:42here's how I access it from a visual interface area. And here's how I access it from the command line. So same thing. I'm just going to open the command line again to show you again. I'm just pressing terminal right here from the Docker manager. What I want to do now is I want to install Claude code on the server,

14:57just to show you what that would look like and make sure it's set up for you. And then we're going to move on to actually setting up Hermes, using it, making connections where it's going to feel a lot more useful. Now I have the command right here, which will install cloud code on the server. You also can just look up cloud code install

15:11and you'll find the install command from the website. It's very easy to see how to do this, and I'm sure many of you already know, but what I can do is I can just paste this in here. Note when you're using this UI, you will need to right click and press paste. The keyboard shortcuts don't work.

15:25So what I'm going to do is just press on enter. And what this will do is prompt me now to sign in with my anthropic or Claude code subscription after obviously it signs in. You also could install codecs. You can install anything you want here, but it's useful to have this AI tool directly on the server because then you can chat with it.

15:41So here, now that I've installed it, I can type Claude. It's going to bring me in here. I'll just go with dark mode. And I will say that I'm going to use my Claude account with subscription to sign in. So from here I'm just going to press on this link, and it's going to open up a new URL where I can sign in and authentic this server.

15:57Okay. So it's given me some code that I'll need to paste here. So I'm going to go back here and I'm just going to right click and press on paste and then paste in that key. And it should then authenticate me here after I press enter. And it looks like we're good to go. So now from here I'll just go.

16:11Yes I trust this folder. And if there's anything that I want it to set up or there was a mistake, I could simply ask Claude Code and it would do that for me so I could say, hey, I'm having trouble getting Hermes set up or selecting my model. Can you help me with this?

16:22And it can then run the commands and do that for you. Now, to get out of this, I'll just hit Ctrl C, but that should be a good kind of failsafe for you in case something goes wrong. All right. So at this point we've installed Hermes. We've set it up. It's on the PS.

16:35We can access the web dashboard as you've seen here. But what we'll want to do now is we want to configure our model if you're using hosting, or it will automatically like give you some AI credits and set something up for you. But it's better if you switch to your own provider. So that's what we'll want to do now. Now there's a bunch of different options for like models or providers

16:54that you can use. You can use again the hosting or one you can use open road or you can use nous portal. You can use Claude, you can use local models. I don't care which one you use. In my case, I will use my ChatGPT subscription because that is one of the cheapest and just best models that you can get,

17:09at least in terms of paying per token, for example, for something else. So here's how you would set up your model. Now from the command line again, you're going to be inside of this Docker container here. What you're going to want to type is Hermes setup. Now if you type Hermes setup this will bring you through a full long

17:26kind of setup flow. But rather than just doing that you can type Hermes setup model. And what this will do is directly bring you to a setup page for choosing your LM provider. Now again, we need a model for Hermes to work properly again. You can see by default it's using this hosting or router, but I'm going to change it over or add another provider and I'm going to use

17:46my OpenAI subscription if I can find it here. See I'm just going to scroll up and we should see a OpenAI one right here. It can be a little bit tricky to find. I'm going to press on enter, and I'm going to use my ChatGPT or Codex subscription again. To do that, I will just copy this code by right clicking and pressing copy.

18:06And I will go to this auth page and paste it in okay. So it says I've signed into Codex. Looks like we're good to go here for the model I'll just go with 6.1 solve for right now for the reasoning level I'll go with medium. And then that should have set up my model for me.

18:21So now if I go back to the web dashboard here and I just create a new chat, okay. So we can see it took a second, but it did update the model here to GPT 6.1 soul. And if we just want to test if Hermes is working, we can just simply type hello. And we should get some kind of response back.

18:37And you can see, hi, Hermes agent. I'm working. Whatever. Okay. And if we press here on the model, we can see we can select all of these different models from this provider right here and view all the other providers. So again first step is set up your model. Choose which one you want to use.

18:51You can use open root or elms to whatever right. There's so many different providers, but I recommend using something from a subscription like your OpenAI or ChatGPT subscription, because that's going to be pretty decent value for what this agent needs. Okay, so we're going to start going through a lot of other stuff, I promise.

19:06We're going to get into skills, memory, all of the Hermes concepts. But right now I do want to show you how to connect to the desktop application. And it's a full application you can run on your computer that either can set Hermes up for you or connect to an existing Hermes installation. So what we're going to do is go to this web page right here. I'll leave it in the description and simply download

19:26this for whatever operating system you're on. Now, once you download it and you install it, it should look something like this. Okay. So on the left hand side you'll be able to see different agents or gateways that you're connected to. You'll be able to see sessions. You can manage bots. And just so you understand the way that this works is

19:43this can give you its own Hermes agent that's running locally on your computer. Or you can connect to something like, we just deployed where we're running one on a virtual private server. Again, I always recommend having it on the PS, even though this can set it up easily on your own computer. So don't mix those two things up. If you just start chatting with this immediately, you're going to be using a different

20:04Hermes agent that's only available when your computer is on. That's not going to be synced with like other versions, for example. And that again has access to all of your files, your system, which you really don't want. So what we want to do is we want to connect this desktop application to the Hermes agent we just deploy.

20:21Now to do that is fairly straightforward. We're going to search for this. I'm trying to zoom in here but it's not really doing too much for me. We're going to search for this settings menu here in the top hand corner. And from here we are going to go and we're going to find gateways.

20:36Now gateways should be here kind of near the bottom of this list. And from here what we're going to do is we're going to press on remote gateway and we're going to put in the URL. I'm going to change this here to the Hermes application that we just created. So again we're going to settings menu here.

20:53We're going over to gateways. From gateways we're going to press on Remote Gateway. And we're going to put in the URL here which is the same as our application URL. So to find this we can go back to our hosting or and we can simply open the Hermes application. From here we can just copy this URL minus the slash okay.

21:13And then go back to the desktop application and paste that in like so. Now from here it's going to say that it uses a username and password to sign in. Or you can press test or something. And it's going to say hey you need to authenticate. So from here we're going to go sign in. And then what we're going to do

21:28is put in the same username and password we had before. Now again if you lost that password you can simply go to the environment here. So we go manage for the Docker container go to the environment tab which I'm going to have to find. And then we're just going to copy the admin password okay.

21:45And then from here we paste this in and sign in okay. Now if we come back to Hermes here we should see something that says save and reconnect and that we were signed in successfully. So we'll go ahead and press on that. And it should reconnect us now and we should be good to go.

22:00It just needs to restart the gateway. Okay. So let's pop this open. And I looks like we are good to go. So we should be able to see. Now GPT 6.1 is our model. Because we're connected to our ChatGPT subscription. We can create new sessions here. View capabilities, messaging, all this stuff that we're going to look at in a second. But from this point forward we're going to use the desktop application,

22:20because I believe this is the best way to interact with Hermes. And then again, we can type something like hello. And assuming that we get a response back here, we're good to go. And you can see the previous chat that we had where we typed this from the web UI. They're all showing up and it's all saved here. Okay. So now that we're in the desktop app, I want to start talking about memory

22:37skills, curator bots, and all of the fun stuff that you need to know about Hermes. Now let's start with memory. Now, Hermes will automatically manage and save memory about you and about things that you ask it to do. There's two files in particular that you want to be aware of, but Hermes will create a lot of other files as well.

22:56Now the first file is called user. And by the way the MD stands for markdown. H is a file format that allows particular AI models to write kind of text. That's very easy for them to parse and understand. Now, this user file is global, which means it's available to all of the bots and everything you're going to see in Hermes in a second.

23:15And basically it will note down information about you. So your name, your job, how you like your answers, whatever. And Hermes will automatically create this for you and update it, but you can also manually tell it to do that as well. The next file is memory MD. Now there can be multiple of these files in different directories, but typically there'll be one memory file that Hermes references

23:35that contains its own nodes. So things like your projects, your setup and different long term information that it's learned. Now it will load these files anytime you start chatting with it. So it knows about you and it knows kind of this long term memory. Now Hermes will also save memory that's not loaded persistently into other files that it can search for later. I'm not going to get into that system too much, but just understand that

23:58while you chat with this, typically it is saving and updating these types of files. Okay, now for example, if we copy this prompt, I'm going to show it to you in a second here and paste it into Hermes. We can see that it will actually start updating these files for us. So I'm just going to paste this in.

24:14Hey, I'm Tim, I run a YouTube channel about AI encoding. I'm in Dubai. Keep answers short with bullet points, whatever. And you'll notice that it tells us it's saving this to memory here. And it's remembered this. Now, if we want to verify that, we can press this button right here in the bottom left hand corner for the command center of this application.

24:32Again, this is from the desktop app which is new. If you press this and you go to maintenance mode at the bottom left here, you'll be able to see a user profile file as well as a memory file. Now the user profile file, it's not showing me the the view of it. I'll show you how to look at in a second, but notice that it has content inside of it, and the reason it has content inside of it

24:54is because I just told it to save something about me. So it's updated that so we can see it. Okay. You also see there's a few other things like running a doctor security audit, backing up your information, whatever. So it's kind of useful to be able to see like your usage, the system, the various sessions and the maintenance

25:10from that command center that I just clicked into. Okay. Anyways, I'm going to say can you tell me the content of user. Okay. Now if I do that, it should just spit out the contents for me so that I can see it and it's going to search those files and hopefully tell me what it is

25:27and we should see that it's updated that info. Okay. So you can see user the file is talking about stores this users Tim runs YouTube channel but AI encoding is in Dubai time prefers this whatever. And we can keep updating this file over time and Hermes will handle that for us. Okay, now the next thing to talk about here

25:43is our personality with the soul.md file. I'm going to copy this prompt. I'm going to explain to you what this is now. Hermes agent or the Hermes agent you're chatting with has a personality. It's how it replies, what it sounds like, who it acts like, whatever. And by default, this personality is stored in a file called soul MD. Now you can view this soul file, which again, I'm going to show

26:05you how to do it in a second, but importantly you can modify it. This allows you to change how Hermes responds, acts, and replies to you. And you can think of it like a system prompt that Hermes will always reference before it sends you a message. So even if you change the AI model, they're always going to have

26:21kind of that similar personality. Now these soul files Hermes has a new update now where you can actually create multiple of them, and it allows you to create multiple different versions of your Hermes agent or have these different bots, which we'll look at in a second in bot mode. So just understand soul defines the personality, and I want to show you how to look at it. Now to look at it, what you're going to do is find these three dots here

26:44where it says manage profile. Okay. Now when you do that it's going to bring up a default profile, which is the one that we're using right now. And you'll notice that there is a soul file that has some default content inside of it. We can read through it. I'm not going to go through all of it.

26:58But basically this is the current personality of Hermes. Okay. So what we can do is we can manually adjust this soul file if we want. We also can create different profiles, which we'll look at in a second, and then load these various profiles to have different personalities for our Hermes agents or bots. So I'm going to leave this as is, but I'm going to tell Hermes agent to rewrite. It's sold to change its name, to be my operations assistant,

27:24to be direct and never send another person without asking me first. Okay, whatever. We're going to go ahead and press on, enter here, and then we should see that it updates that file for us, and we'll be able to review that by viewing the profile, and that it should change materially the way that this replies to us.

27:40Okay. So you can see that it rewrote the soul file. If I go here to the profile, this is now the new profile, right. The new soul file in this markdown format. And I say, hey, what is your name? It should tell me that its name is Atlas here, because that's what I've told it.

27:55Say Atlas, my operations assistant. And by the way, if you guys are wondering how I'm actually communicating here with Hermes because I'm using my voice, I'm using a voice dictation tool here called Whisper Flow. Basically, this is speech to text. It's one of the fastest, best ways to talk with things like AI models.

28:12I use it pretty much every single day. You can see I have over 292,000 words. I highly recommend that you guys install a tool like this. I have a partnership whisper, so I'll leave a link to it down below in case you guys want to use this yourself. But just again, I use it all the time.

28:26It automatically formats your text. It's very, very good. So I would highly recommend. Okay, anyways that is sold. So hopefully you kind of understand that and how that changes the personality of the model. And now I want to move on to skills. Now basically Hermes will automatically save things as skills like we discussed when you ask it to do something this way becomes more efficient over time

28:47and uses less tokens because it can follow a procedure that it's already learned. Now you also can tell it to learn something manually using the slash learn command. This allows you to invoke it to create the skill for you. You also can manually tell it to make a skill, but slash learn will kind of follow a process that it has. So for example, if I just asked Hermes to make a YouTube brief for me, then I could say something like slash, learn how you just made the YouTube brief,

29:13and then it will learn how to do that and create a skill. So I want to show you kind of how that works with the skill process. So I'm going to say something like this. Hey, can you make a YouTube brief for me that just quickly explains what Hermes agent is. Basically, I want to make a five minute long YouTube video that has a few different sections

29:30that goes over Hermes agents, its usefulness, how you install it, whatever. Keep it simple to the point and make sure it's adjusted for something like tech with Tim, which is my YouTube channel. I'm just kind of rambling some random stuff into Hermes here, but I want to show you the point that it's going to attempt this task, and then we can force it to actually

29:48create a skill based on that and to update and improve that over time. Okay. So it just created this for me. And what I'm going to do now is type slash learn, okay. And I'm just going to do space here. And I'm going to tell it create a skill or actually let's not go create a skill.

30:04Let's go. How you just created this YouTube brief. So I'm saying slash learn how you created the YouTube brief. And now what it's going to do is learn this and create a new skill for me. Okay. So you can see now what it says is that it's updated this YouTube content inside of media. Now it looks like what actually occurred here is there was already some skill related to

30:23generating YouTube content. So it just simply updated that for us. But if we want to view that and verify it, we can go to capabilities. And we'll see here that there's actually this media skill right, called YouTube content. And I can turn it on or off. And if we scroll through here, we can see exactly what this looks like.

30:39Right. And how we create a brief, how we run it, you know, whatever. And it said, hey, make a five minute YouTube brief, outline a tech with ten video. So it updated. This right is learning how we're using these skills. Now there's a bunch of other skills that come out of the box that again,

30:54it can update for you. And you also can install other skills by kind of scrolling through the installable ones here. And also, if I can bring this up, I got to find the pull here. There is a skills hub that you can directly install skills from. Now. Be careful with these and you should read them before

31:12you install them, because some of them can be malicious. But if I click into like Apple reminders, I can add it to this agent and directly install it. Okay, so there's all these different skills and that's really the bread and butter of Hermes skills. Managing skills, adding skills. You get the idea. So if you want to have so you'll definitely want to have a look at these, manage them and also disable

31:32any of the ones that you don't want to use inside of this particular agent. Okay. So let's go back to our session here and continue. So as you saw there are these commands that you can run okay. Now there are a lot of commands that I cannot possibly go through all of them. What I would recommend you do though is type slash and just have a look

31:50at what's available here. Because oftentimes you will want to use some of these commands. Now for example, if you want to reference a skill directly you can just type the name of that skill. So like YouTube content. Right. And then just put a space so YouTube content and then tell it what it is that you want to generate.

32:06That's probably the most basic usage of commands. Next would be something like slash new. This will create a new session for you. So it's the same thing as just pressing new session effectively. Some other commands. Slash save. You can save the current session transcript to a file, which is pretty useful. Slash retry straight forward.

32:23Retry the previous thing. Now there's some other commands like slash title changing the session title hand off. Not going to go through this right now, but you can hand off this session to another gateway or another platform compress. This allows you to compress the current context. So it will summarize everything that's currently here.

32:41And then kind of give you a cleaner context with less information. Rollback. Go back to a previous system slash stop slash background right. Run a prompt in the background slash. By the way you can ask a side question. Well something's going on. So there's a lot of commands again that you can have a look at here I don't want to go through all of them.

32:58But you'll see them here. And there's some fun ones like for example adding a pet. So you go pet. So with the pet you'll see it gives you some options here list scale off toggle whatever. I'm not going to go through them. Point is that if you click into those commands there, they will actually show you the options of what to press next. And if you're unaware of how to use them, you can just ask Hermes

33:16and it will tell you exactly what you need to pass. Okay, so now that we've looked at some commands and hopefully you understand these skills, let's quickly look at tools. If you go to capabilities and you go to tools, you're going to see that there are sets of tools like file operations, terminal processes,

33:32code execution, memory, whatever that you can enable and disabled. Now generally you're going to want to leave all of these on for your default Hermes agent, but you may want to create subjects or different profiles of agents that, for example, are read only so they can't accidentally make a mistake. You may want to have ones

33:49that don't save something to memory, or they don't use your existing skills. Or maybe they don't want to use a cron job, or you get the idea. So you can toggle these on and off and adjust kind of what tools the agents have access to. As you can see right here. Okay. Now next you have connectors. Again, we kind of already briefly went through this, but this is a new feature that Hermes has that allows it to natively connect to a ton of different applications,

34:12like asana, for example, or Canva or Cloudflare. So I'm not going to go through the connections because it really depends on what you want to use. But rather than having to install your own applications, which of course you can do now, Hermes has a bunch of popular ones that are built in that makes your life a lot easier.

34:28Next, you have plugins. Now, plugins are typically sets of things like skills as well as tools and potentially other connectors. You can install these plugins directly from the plugin catalog. And again, the plugins are kind of a little bit more comprehensive than skills, where they typically include more things bundled together. Now you will only need plugins for certain operations, right? And you can install them directly from git or from the plugin catalog down here,

34:52like a memory wiki, right? A dashboard reset watch quota, whatever. Okay. And you can check them all out. I'm not going to go through them, but just be aware that they exist. Okay. So now that we've looked at that, let's look at the messaging platforms and how you can use this from something like your phone. Now again, I'm not going to really go through this

35:09because it matters what one you want to use. And I'm not going to show you to set all of them up. But right now we're using this from the desktop app, which is what I would recommend from your computer. But if you want to chat with Hermes agent from your phone, you can do that. And you would typically do that by connecting WhatsApp or Telegram or Discord or Slack or something else

35:26where you can chat with it in a different gateway or interface. To do that, you can just go to messaging here in the desktop application, if you go to something like telegram, you can literally just scan a QR code and get it set up and chatting with it right away. If you go to discord, it's going to tell you you need to get some discord credentials and put in your bot token

35:44and then say what users can chat with it, add it to a discord server. Get the idea? Slack? Same thing. So go through this if you do want to chat from your phone or from another platform. For now, we'll just leave it in the desktop app. Okay, so now I want to get to one of the new features of Hermes, which is bots.

36:00Now, as you already saw, right? We can just directly chat with the agent. We can make new sessions, we can give it skills, it can use our connectors. And I mean, you guys know what you use AI agents for, right? I'm not going to try to give you a bunch of random use cases here that aren't going to be applicable.

36:13However, this is a new feature that recently came out called bots. Now, this allows us to create different versions of Hermes agents effectively that we can chat with that have different configuration, different tools, different connectors and different soul MDS. So what I can do is go to this bots section right here, press this plus button and I can make a new bot. Now when I make the new bot,

36:35I need to specify where I'm making it, so I'm going to make it on this. Notice I have multiple books. I was playing with it before, and for the bot you can give it an image or like an icon. You can also then give it a name, a title, a description. And if you go to advanced you can modify.

36:49It's sold on. You can change the model that you want to use. So I can say, hey, I'm going to go, I'm going to use ChatGPT subscription. And for this one maybe I want to use Luna, right. Because I don't want to use so many credits and I want this to be a little bit cheaper.

37:03So maybe for this one I'm going to call this life coach. Okay. And for the title of this same thing, we'll just go life coach. And I'm going to say this should help me with my life and generally speaking, guide me kind of like a therapist or tell me what direction I should go in based on the struggles I'm facing. I'm just doing something random that's going to be like unique to understand for sold, I want you to be understanding, caring,

37:29compassionate, and really asking me good, hard hitting questions. Don't just take what I say for granted. I want you to kind of always quiz me and make sure that I'm giving you real answers, okay? And then we can go ahead and just create this button. Of course, there's other settings you can adjust, but what will happen

37:45now is that that bot should appear here in one second in this left hand side. Okay. So the bots created it. Just took a second here. I had to refresh my window here. But what I'm going to do is just paste this. Hey, I'm really struggling with my career. I'm thinking about quitting YouTube.

37:59What should I do? And I just want to show you that we're going to get a different type of response here. It's going to be very fast because we're using the other model based on that. Solved. Now you notice we have the default Hermes agent, which I can chat with here, give it its own sessions. All I have life coach and I'm kind of in these like DMs with these different bots,

38:15which is super useful. And then I also can make a group chat. So if I want, I can add Life Coach and Hermes into this chat and create a group with them, and then they can start messing with each other and I can also message with them. So I'm just going to say at all.

38:30How is it going? And now both of these bots ideally should reply so we can see live coaches giving us a reply. And now Hermes is thinking and I can have these different bots with different tools, different connections and different goals, all working together in one workspace, which is pretty interesting. And you'll see that they can now just start chatting with each other

38:49autonomously, without me actually needing to give them a reply. Now, I don't know exactly how this works, so I haven't played with it too much. But notice that, like they're just going in through this conversation and like they've chatted multiple times now and I can tell them to talk with each other. So I'm going to say for 30s just talk back and forth with one another. And I mean, let's see actually, if that works and if they start chatting

39:11with each other. Okay. So you can see now they're kind of going back and forth in this turn by turn conversation. I don't really know when they're going to stop or if they can keep track of time, but I mean, it looks pretty cool that we can set that up. And that's a really interesting feature that Hermes recently

39:26added, which now means I don't need to make like five different Hermes agents. I can just set up these group chats, I can set up different profiles, and then I can tag them and communicate with them directly, either in a group chat or in their own sessions. All right. So let's quickly just look at some other stuff. If we go into one of the bots or the agents here, what I can do is I can show the right sidebar, and it's going to show me a full file

39:47view of every file that these agents have access to. So, for example, if we look at the sole notice that this one is different here, right, than it is for this Hermes agent here, where if I go over to soul, you can see that it looks different. You're also going to notice that they have their different files.

40:03So they're all in their own kind of separate workspace here. And these are independent bots that have their own memory, own information, own session log and all of that info. So that's definitely worth noting. Now let's go back to Life coach for one second. Another thing we can do, let's move this back is view their scheduled jobs.

40:19I'm just going to close this here. So schedule jobs are things that are going to run recurrently. Now these are what I talked to before called Kranz. You can manually create a cron by pressing this new cron button here and set when to run it. Give it a time, a name, prompt, whatever and then it will go ahead and run.

40:36Or you can just tell it to set it up. So I can say every morning at 9 a.m. Dubai time, can you just greet me and ask me how my days going? Okay, so this would be a scheduled job and then it would automatically do that and greet me. And I could also tell it where to greet me

40:50if I set up a messaging platform like telegram or something, for example. Now you're also going to see this screen thing. If we install this, it will let us view what the agent's actually doing. If it's viewing a terminal or something. I'm not sure exactly how that's going to work on the virtual private server,

41:04so I'll just leave it uninstalled for right now. But you'll notice that there's a bunch of other settings like individual settings, right? A layout editor. We can view a HUD mode. Which brings us to this weird UI where we have this kind of small window that we can leave open and I guess move around on our screen.

41:20I'll go back to this. And now if we kind of close the schedule jobs down and review it, you can see that we have this Dubai cron that has been automatically created. Okay. Now there's obviously a lot more that we can do here with Hermes agent. But I think at this point I've given you a pretty solid grasp of how to set it up, specifically the new features like the bot mode, which I think is one of the most interesting things

41:41that Hermes has created. The group chats, the soul files, the memory, how you view it, how you understand this interface, how you deploy it. And I am going to leave the video here. If you guys want a more advanced guide going through more particular use cases, then definitely let me know in the comments down below.

41:56With that said, I hope you guys enjoyed. If you did leave a like subscribe and I will see you in the next one.