{
  "video_id": "D3dQqqDx2V4",
  "segment_count": 475,
  "duration": "14:30",
  "full_text": "Hermes has the best memory of any AI agent. It remembers your conversations and gets smarter the more that you use it. But, there's one thing it can't do. And once you fix it, you give Hermes superpowers and unlock completely new capabilities. So, in this video, I'll show you exactly how to fix that [music] to give Hermes a self-improving knowledge base based on Andre Karpathy's LLM principles. It'll save you hours of time and get you light-years ahead of everybody else. If you're new, I'm Jack. I built an automatic startup with a gazillion customers, and now I run my own AI businesses. And here, I just share the stuff >> [music] >> that actually works. So, if you haven't already, grab that beautiful coffee, and let's dive straight in. Now, most LLMs feel like they have amnesia. They just forget things randomly. I can mean and they don't even know that they've forgotten them. This is why Hermes is so good. And before you can understand its limitation and how this actually adds value. And if you stick with this video to the end, you will have new capabilities in your Hermes agent that will feel completely different and get you further ahead. But, to do that, we have to understand what is Hermes memory and how does it work. So, the best way to think about it is that Hermes gets better the more that you physically use it. It has this loop that is like the top-level way of doing it. And I've I've drawn some things you can take pause and have a look at if you want to. Fundamentally though, you have a conversation with Hermes agent, it writes a note that goes into memory.md or user.md, and then it recalls it when you're having conversation. This works on notices, you know, pulls durable facts, outbound or internal jump. It works on files, and it can also recall and search those things before it answers you. But, there's one catch, one limitation to this Hermes system. And that's the fact that it knows only what you've said. So, your inbox, your calls, your docs, your research, these beautiful YouTubers you followed, all this great knowledge is invisible to Hermes. And that is blind spot. It knows you super well, and I freaking love Hermes. I talk about Hermes all the time, but it doesn't have all of this additional context. It doesn't know, you know, if you're trying to grow on Instagram or YouTube, or you're trying to scale your email marketing system, it doesn't have that knowledge yet. It knows you very well, but doesn't have everything to the right-hand side of the wall. If you think about it this way, Hermes knows you, but it doesn't know your email threads, meeting notes, or notebook LM. It can, of course, call those when it's asked to palm, but it's not in it if you like it's conversational memory in that sense. Said another way, Hermes knows you, it doesn't know your inbox. And so, the idea to solve this is we want a knowledge base that grows by itself, that we can store all of this information, that Hermes can call and reference whenever it wants to, and it can also do very interesting capabilities that you've not seen discussed anywhere on this particular topic, and I'll show you what I mean by that. So, if you look at Andrej Karpathy, by the way, co-founder of OpenAI, if you don't know who I tell you are, he led AI at Tesla, he's a fan of Eureka Labs, and he even coined the expression vibe coder, okay? So, he's pretty well known in the space. Now, his idea here was an LLM wiki, you might have seen it as referred to as Obsidian Rag, and its core idea with Hermes Agent is that it rewrites itself as it grows. So, we can take all of the information that we want to, right? Like, again, expert commentary, things that we build in notebook LM, and effectively, what it will do is the more we add to it, the better it gets. As you add it, it checks linkages, it fact-checks things, and it basically creates an ever-growing self-referential, almost Wikipedia of knowledge for anything that you want to learn about or any historical context you have. you think about it like this, Hermes knows you, the wiki knows your world, and now when we combine this with Hermes Agent, your agent can read both and basically do very cool things when we wire these two things together, your world and you. Think about it from that point of view. So, now we understand the limitations of Hermes memory system, the first thing we need to do is get this beautiful LLM wiki set up so we can actually reference this within Hermes and expand its working knowledge so it can answer questions about essentially anything that you want. And by the way, once it's set up, you can even visualize and see your entire LLM Wiki Obsidian system all within your dashboard. It is so freaking cool. So, the first step we do is head over to this LLM Wiki. Effectively, what it is is explaining in detail basically architecture works, how the system works, and it's got all of the detail here, which is freaking awesome. So, all you're going to literally do is come down and grab this URL. Then you can open up your language model of choice. You can do this in Hermes, but we've just had Opus 4.8 drop, which is the world's most capable model as of today. I've got some interesting thoughts on that. So, we can use the OAuth inside the Claude app. What that basically means is we don't have to spend any API credits if we're just using the app and you have the subscription. Alternatively, you can just open up Hermes and it can follow these exact same instructions without a problem. But for the purposes of this, what I'm going to do is basically say, \"Hey there, I want you to familiarize yourself with the idea in this URL, and I want you to create for me a desktop folder with all the requisite and let me know once that is complete.\" And then we come down and we simply paste this one in here, and effectively, it will kind of auto set up based on the instructions that exist there. Now, since I already have one set up, it's found that I saved it as Obsidian Wiki, and effectively, what it will do is build out a structure that looks like this. Just so you can visualize it a little bit easier, I'm going to pull up the anti-gravity IDE, not 2.0 I might add, it's anti-gravity IDE, just so you understand what this physically looks like. And as you can see, I've added various little things in here like transcripts of various YouTube videos I think are interesting about, you know, different strategies for how to make better content and that kind of thing, and all of it is within there itself. And you can ask it questions if you want to. So, for example, if I'm querying this Wiki, one of the things I might do, and as you can see down here, if you come to claw.md, you can see it's got the whole structure outlaid here, and I'll put this claw.md down below for you as well so you can use this as an additional thing. But you can see it's got an injection workflow, so when a new file is added, it's going to read it, it's going to discuss it, it's going to write the source page, okay, and it's going to update any affected pages and essentially flag any contradictions. So, the idea is like if you add a new piece of knowledge and it contradicts with something else, we identify those contradictions and the knowledge base gets stronger over time. So, then I might say something like, \"Hey there, dude. Could you just give me two quick uh tips for making better intros?\" Okay, and it can actually go ahead and search the Obsidian Wiki. It works exactly the same in Claude Code. By the way, if this all sounds like I'm speaking Mandarin, you can watch this video here for a full breakdown on exactly what it all means to build in Claude Code and Anti-Gravity. But, for the purposes of this, I'm basically just explaining how it's all connected. Again, you can also do this in Hermes Agent. It will set everything up for you. I'm just showing you in this environment so you can see it visually. And as you can see, drawing a YouTube content strategy synthesis in your Wiki, here two quickly actionable tips for loving the intros. There you go. And it's basically found this info that information from our Wiki. Now, the cool thing here is that we can also ingest anything that we want to. But, I'm going to show you exactly actually how you can take this to your novel by essentially doing all of this within Hermes. But, to do that, the first thing we have to do is basically connect this file this desktop where your personal Wiki's going to live. And we need to share that with Hermes. So, if you're using this dashboard, all you're going to do is literally come down here and you'll see the instructions on the left-hand side here for the Obsidian Wiki. We basically you just disconnect this and it will you can just enter in the location of where your vault is and then literally just copy this information here. Now, if you don't have, okay, the actual um location, you can just say to Claude or anyone on your laptop, \"Hey, what is the URL location of this?\" And once you've done that, you can then connect it. So, for example here, I'll just come and say, \"Hey there, I would like you to reference my basically my LLM Wiki, my Obsidian Wiki.\" Effectively, this is going to have information that is external to you. So, you know everything about me. This is going to be a source of information knowledge that you can query in our conversations to better answer my questions. This is going to be things like my meeting transcripts, external insights from experts that are going to help you basically make better decisions. Okay. And what I think we can do here, if you haven't already, and I recommend you do this, is build out what we call like a kind of LLM wiki skill or an Obsidian skill. And effectively, you can do this if using personas here, that's freaking awesome. I've covered that in depth on the channel how you can just basically add in anything you want to here. For instance, by clicking on persona, coming down, picking the person, naming it here, and this would be something like, I don't know, LLM wiki. And here would be use this skill when answering any questions about strategy or any contacts about meetings, notes, that kind of thing. And then just basically build out a system prompt. And then you can pick the model that you want to use to build that out. Now, interestingly here, you can actually then, if you just go ahead and update this, and give this prompt over to Hermes, it will be fully up to date. You can also, within Hermes, if you want to, just effectively ask it build out that skill. I just like to do this personally cuz I like to visualize and see all the skills, so I know exactly what I want to improve and how I want to physically go about it. So, for example, if I just click a feature, I can come down and test this. Either basically go to my LLM wiki and just tell me, for example, three tips on how to make a better intro. Okay, so we can do this, give that to Hermes. And now what Hermes is doing, crucially, is consulting this ever-growing, self-referential, self-improving database of knowledge. Here you go, you come down here, and it's basically consulted this, and I can say, \"Hey, can I just confirm where did you get this information from?\" And as you can see, I already can see here, it read the file exactly where my Obsidian wiki is, and it found the information from there. And look, it's gone ahead and grabbed it. So, effectively, what we've done now is we've got our own individual LLM wiki that contains everything about all the calls that we've had, maybe it contains things about knowledge from experts, and then we have Hermes' own internal memory system, which is by itself incredible, and we combine these two things. Now, to build out your personal LLM wiki, when would you use that? Well, let me give you an example. Let's say for example, you are searching internet and you come across a website like this. This is high agency in 30 minutes, which is a beautiful article by the way, highly recommend it. What you can do now is basically have this index. So, let's say for example that you really like the principles of high agency and what it means to be high agency and how much very valuable. And you kind of want Hermes to understand this stuff when you're talking to it. Like maybe you want to grow on LinkedIn and therefore you want a LinkedIn master article to be in Hermes' memory. Well, what we can do really now is we can copy the text or copy the article. We can go straight over to Hermes. Essentially, what we can do guys is drop it in and be like, \"Hey there, I want you to go ahead and basically index this article into my Obsidian wiki.\" Which means then that it will reference this and understand this whenever we physically want to go ahead and use it. And look, it will just gone down, it's checked it out. This is George Mack's high agency in 30 minutes. Quick tell the offer on the page. Before I create the source page, what do you want emphasized? So, I'm going to say, \"What I'd like to do, bro, is basically embed it into the wiki in line with the principles set out in that claw.md and just kind of index it as part of that. And then from that, it's actually gone ahead and completely indexed it. So, I can now ask it questions about high agency and I'll do that. For example, what are the criteria of a high agency person? Okay, I'm going to send this one off and it will now reference that Obsidian wiki to pull that information. As you can see, high agency people are spotted by four main signals, where teenage hobbies, treadmill energy, unpredictable opinions, blah blah blah. So, the idea here is if you're just talking to Hermes agent, it will remember stuff. Remember, Hermes' memory is incredible. You can even ask questions, \"What did I talk to you about on the 15th of April?\" And it will be able to pull those specific things. The difference here with the wiki is this is an very large knowledge base where we're indexing it. It's a growing corpus of knowledge of things that you find interesting from experts which Hermes can now reach over, grab, and access. And the cool thing is if you're ever working in cloud code, has access to the same knowledge. So, I can say, \"Hey there, do me a favor. What are the spokes of a high agency person? This will will visible in my Obsidian wiki.\" And as you can see, it's on the scale Obsidian ask. It comes on its founder, which is awesome. And look, it's giving us all the information that spoke to me I'm thinking about blah blah blah. So, now essentially Claude and Hermes' agent have access to the same memory. And this is part of the idea of the operating system that we universalize this knowledge and everything's connected in one system rather than having a thousand different things and a thousand different interfaces to use. And so, one of the other cool things that we can do here and if you think about what we've done here, we've wired the memory both ways. So, we can access this LLM Wiki and we can auto ingest files and data into this Wiki without polluting the memory of Hermes' agent, which I'm going to show you in a second. It's very, very cool. And we have this beautiful two-way system giving Hermes this kind of access to this whole new external system, external knowledge. And so, if you think about it, we've got this great bidirectional relationship. Hermes knows everything about you and then we have this LLM Wiki where Hermes itself can just send things over there. Like, if you have a really big conversation and you really want to crystallize that knowledge, you can essentially just say, \"Hey, index this into my Wiki.\" So, it grows this growing corpus of knowledge. Equally, anything from calendars, Gmail, meetings, we can just set up auto basically automatic cron jobs, also known as just like things that run in the background to save that. And we can do that all within Hermes' agent. For example, I could say, \"Hey, what I'd like like you to do is set up a recurring task on every day that goes through to see if I had any new meetings that are shown in Granola. And if I have, I'd like it to add those, please, to my Wiki under kind of like meeting notes, meeting information. Okay? Go ahead and save that one off above, which is really cool. Now, Granola is just the app that I'm using to basically record meetings that I go into. They're not like a sponsor of the video or anything like that, but they're really cool cuz it just basically lets you chat to all of your meetings. And now I'm going to have this database, all this knowledge I can just reference whenever I need to. And there you go, asking us a question. So, I could say, \"Hey, why don't you go ahead and just run it at 9:00 a.m. every morning?\" And I'll say, \"That sounds like a great place to store it. Just add it wherever you think would be most appropriate inside that database.\" And just like that, it's now created a job. So, essentially, every single day now, that will run, it will check all my meetings, if I've got any meetings, it's now added to my LLM Wiki, which is Cap's whole memory system in a nutshell. And then we have access to that in Hermes whenever we want to. And the coolest thing about this, guys, is you could do this with Notebooks. Like, you can connect Notebook LLM to Hermes Agent. I'll put a video on screen if you haven't seen that. Or Cloud Code, get like, you know, 50 to 100 different areas of like expert knowledge from Google's number one research and intelligence platform, I can ingest it in my local Wiki on my computer, and then Hermes can just reference that whenever I want to with this skill. And Hermes itself can create automations to ingest it, so your knowledge base just grows exponentially and will just never forget the stuff that it needs to know. Now, memory is great, but it's only one part of the puzzle. You need to understand how to leverage all of the different aspects of Hermes if you want to get the full capabilities. So, the next thing I'm going to do is learn these capabilities by watching this video right here.",
  "timestamped_text": "0:00 Hermes has the best memory of any AI\n0:03 agent. It remembers your conversations\n0:05 and gets smarter the more that you use\n0:07 it. But, there's one thing it can't do.\n0:09 And once you fix it, you give Hermes\n0:12 superpowers and unlock completely new\n0:14 capabilities. So, in this video, I'll\n0:16 show you exactly how to fix that [music]\n0:18 to give Hermes a self-improving\n0:20 knowledge base based on Andre Karpathy's\n0:24 LLM principles. It'll save you hours of\n0:26 time and get you light-years ahead of\n0:29 everybody else. If you're new, I'm Jack.\n0:30 I built an automatic startup with a\n0:32 gazillion customers, and now I run my\n0:34 own AI businesses. And here, I just\n0:36 share the stuff\n0:37 >> [music]\n0:37 >> that actually works. So, if you haven't\n0:39 already, grab that beautiful coffee, and\n0:41 let's dive straight in. Now, most LLMs\n0:43 feel like they have amnesia. They just\n0:45 forget things randomly. I can mean and\n0:47 they don't even know that they've\n0:48 forgotten them. This is why Hermes is so\n0:50 good. And before you can understand its\n0:51 limitation and how this actually adds\n0:53 value. And if you stick with this video\n0:55 to the end, you will have new\n0:57 capabilities in your Hermes agent that\n0:59 will feel completely different and get\n1:01 you further ahead. But, to do that, we\n1:03 have to understand what is Hermes memory\n1:05 and how does it work. So, the best way\n1:07 to think about it is that Hermes gets\n1:09 better the more that you physically use\n1:12 it. It has this loop that is like the\n1:14 top-level way of doing it. And I've I've\n1:15 drawn some things you can take pause and\n1:17 have a look at if you want to.\n1:18 Fundamentally though, you have a\n1:19 conversation with Hermes agent, it\n1:21 writes a note that goes into memory.md\n1:23 or user.md, and then it recalls it when\n1:26 you're having conversation. This works\n1:27 on notices, you know, pulls durable\n1:30 facts, outbound or internal jump. It\n1:31 works on files, and it can also recall\n1:33 and search those things before it\n1:35 answers you. But, there's one catch, one\n1:38 limitation to this Hermes system. And\n1:41 that's the fact that it knows only what\n1:43 you've said. So, your inbox, your calls,\n1:45 your docs, your research, these\n1:47 beautiful YouTubers you followed, all\n1:49 this great knowledge is invisible to\n1:52 Hermes. And that is blind spot. It knows\n1:53 you super well, and I freaking love\n1:55 Hermes. I talk about Hermes all the\n1:56 time, but it doesn't have all of this\n1:58 additional context. It doesn't know, you\n2:00 know, if you're trying to grow on\n2:01 Instagram or YouTube, or you're trying\n2:03 to scale your email marketing system, it\n2:05 doesn't have that knowledge yet. It\n2:06 knows you very well, but doesn't have\n2:08 everything to the right-hand side of the\n2:10 wall. If you think about it this way,\n2:12 Hermes knows you, but it doesn't know\n2:14 your email threads, meeting notes, or\n2:15 notebook LM. It can, of course, call\n2:17 those when it's asked to palm, but it's\n2:18 not in it if you like it's\n2:20 conversational memory in that sense.\n2:22 Said another way, Hermes knows you, it\n2:24 doesn't know your inbox. And so, the\n2:25 idea to solve this is we want a\n2:27 knowledge base that grows by itself,\n2:29 that we can store all of this\n2:31 information, that Hermes can call and\n2:32 reference whenever it wants to, and it\n2:34 can also do very interesting\n2:36 capabilities that you've not seen\n2:37 discussed anywhere on this particular\n2:39 topic, and I'll show you what I mean by\n2:40 that. So, if you look at Andrej\n2:42 Karpathy, by the way, co-founder of\n2:43 OpenAI, if you don't know who I tell you\n2:44 are, he led AI at Tesla, he's a fan of\n2:46 Eureka Labs, and he even coined the\n2:48 expression vibe coder, okay? So, he's\n2:50 pretty well known in the space. Now, his\n2:52 idea here was an LLM wiki, you might\n2:54 have seen it as referred to as Obsidian\n2:56 Rag, and its core idea with Hermes Agent\n2:59 is that it rewrites itself as it grows.\n3:01 So, we can take all of the information\n3:03 that we want to, right? Like, again,\n3:04 expert commentary, things that we build\n3:06 in notebook LM, and effectively, what it\n3:08 will do is the more we add to it, the\n3:10 better it gets. As you add it, it checks\n3:12 linkages, it fact-checks things, and it\n3:15 basically creates an ever-growing\n3:17 self-referential,\n3:18 almost Wikipedia of knowledge for\n3:21 anything that you want to learn about or\n3:22 any historical context you have. you\n3:24 think about it like this, Hermes knows\n3:26 you, the wiki knows your world, and now\n3:29 when we combine this with Hermes Agent,\n3:30 your agent can read both and basically\n3:32 do very cool things when we wire these\n3:34 two things together, your world and you.\n3:36 Think about it from that point of view.\n3:37 So, now we understand the limitations of\n3:39 Hermes memory system, the first thing we\n3:42 need to do is get this beautiful LLM\n3:44 wiki set up so we can actually reference\n3:46 this within Hermes and expand its\n3:49 working knowledge so it can answer\n3:50 questions about essentially anything\n3:53 that you want. And by the way, once it's\n3:54 set up, you can even visualize and see\n3:56 your entire LLM Wiki Obsidian system all\n3:59 within your dashboard. It is so freaking\n4:01 cool. So, the first step we do is head\n4:02 over to this LLM Wiki. Effectively, what\n4:04 it is is explaining in detail basically\n4:06 architecture works, how the system\n4:08 works, and it's got all of the detail\n4:10 here, which is freaking awesome. So, all\n4:11 you're going to literally do is come\n4:12 down and grab this URL. Then you can\n4:14 open up your language model of choice.\n4:17 You can do this in Hermes, but we've\n4:18 just had Opus 4.8 drop, which is the\n4:21 world's most capable model as of today.\n4:23 I've got some interesting thoughts on\n4:24 that. So, we can use the OAuth inside\n4:27 the Claude app. What that basically\n4:29 means is we don't have to spend any API\n4:31 credits if we're just using the app and\n4:33 you have the subscription.\n4:35 Alternatively, you can just open up\n4:36 Hermes and it can follow these exact\n4:38 same instructions without a problem. But\n4:41 for the purposes of this, what I'm going\n4:42 to do is basically say, \"Hey there, I\n4:43 want you to familiarize yourself with\n4:46 the idea in this URL, and I want you to\n4:49 create for me a desktop folder with all\n4:51 the requisite and let me know once that\n4:52 is complete.\" And then we come down and\n4:54 we simply paste this one in here, and\n4:55 effectively, it will kind of auto set up\n4:58 based on the instructions that exist\n4:59 there. Now, since I already have one set\n5:01 up, it's found that I saved it as\n5:03 Obsidian Wiki, and effectively, what it\n5:04 will do is build out a structure that\n5:06 looks like this. Just so you can\n5:07 visualize it a little bit easier, I'm\n5:09 going to pull up the anti-gravity IDE,\n5:11 not 2.0 I might add, it's anti-gravity\n5:13 IDE, just so you understand what this\n5:15 physically looks like. And as you can\n5:17 see, I've added various little things in\n5:18 here like transcripts of various YouTube\n5:20 videos I think are interesting about,\n5:22 you know, different strategies for how\n5:24 to make better content and that kind of\n5:25 thing, and all of it is within there\n5:27 itself. And you can ask it questions if\n5:29 you want to. So, for example, if I'm\n5:30 querying this Wiki, one of the things I\n5:32 might do, and as you can see down here,\n5:33 if you come to claw.md, you can see it's\n5:36 got the whole structure outlaid here,\n5:37 and I'll put this claw.md down below for\n5:39 you as well so you can use this as an\n5:40 additional thing. But you can see it's\n5:41 got an injection workflow, so when a new\n5:44 file is added, it's going to read it,\n5:46 it's going to discuss it, it's going to\n5:47 write the source page, okay, and it's\n5:49 going to update any affected pages and\n5:51 essentially flag any contradictions. So,\n5:54 the idea is like if you add a new piece\n5:55 of knowledge and it contradicts with\n5:57 something else, we identify those\n5:58 contradictions and the knowledge base\n6:00 gets stronger over time. So, then I\n6:02 might say something like, \"Hey there,\n6:03 dude. Could you just give me two quick\n6:05 uh tips for making better intros?\" Okay,\n6:07 and it can actually go ahead and search\n6:08 the Obsidian Wiki. It works exactly the\n6:11 same in Claude Code. By the way, if this\n6:13 all sounds like I'm speaking Mandarin,\n6:15 you can watch this video here for a full\n6:16 breakdown on exactly what it all means\n6:18 to build in Claude Code and\n6:19 Anti-Gravity. But, for the purposes of\n6:21 this, I'm basically just explaining how\n6:23 it's all connected. Again, you can also\n6:25 do this in Hermes Agent. It will set\n6:27 everything up for you. I'm just showing\n6:28 you in this environment so you can see\n6:31 it visually. And as you can see, drawing\n6:33 a YouTube content strategy synthesis in\n6:34 your Wiki, here two quickly actionable\n6:36 tips for loving the intros. There you\n6:38 go. And it's basically found this info\n6:40 that information from our Wiki. Now, the\n6:42 cool thing here is that we can also\n6:44 ingest anything that we want to. But,\n6:46 I'm going to show you exactly actually\n6:47 how you can take this to your novel by\n6:49 essentially doing all of this within\n6:51 Hermes. But, to do that, the first thing\n6:53 we have to do is basically connect this\n6:55 file this desktop where your personal\n6:57 Wiki's going to live. And we need to\n6:59 share that with Hermes. So, if you're\n7:01 using this dashboard, all you're going\n7:02 to do is literally come down here and\n7:03 you'll see the instructions on the\n7:05 left-hand side here for the Obsidian\n7:07 Wiki. We basically you just disconnect\n7:09 this and it will you can just enter in\n7:11 the location of where your vault is and\n7:13 then literally just copy this\n7:14 information here. Now, if you don't\n7:16 have, okay, the actual um location, you\n7:19 can just say to Claude or anyone on your\n7:21 laptop, \"Hey, what is the URL location\n7:23 of this?\" And once you've done that, you\n7:25 can then connect it. So, for example\n7:26 here, I'll just come and say, \"Hey\n7:28 there, I would like you to reference my\n7:30 basically my LLM Wiki, my Obsidian\n7:33 Wiki.\" Effectively, this is going to\n7:35 have information that is external to\n7:37 you. So, you know everything about me.\n7:39 This is going to be a source of\n7:40 information knowledge that you can query\n7:42 in our conversations to better answer my\n7:44 questions. This is going to be things\n7:46 like my meeting transcripts, external\n7:48 insights from experts that are going to\n7:50 help you basically make better\n7:52 decisions. Okay. And what I think we can\n7:54 do here, if you haven't already, and I\n7:56 recommend you do this, is build out what\n7:58 we call like a kind of LLM wiki skill or\n8:00 an Obsidian skill. And effectively, you\n8:02 can do this if using personas here,\n8:03 that's freaking awesome. I've covered\n8:05 that in depth on the channel how you can\n8:06 just basically add in anything you want\n8:08 to here. For instance, by clicking on\n8:09 persona, coming down, picking the\n8:11 person, naming it here, and this would\n8:12 be something like, I don't know, LLM\n8:14 wiki. And here would be use this skill\n8:16 when answering any questions about\n8:18 strategy or any contacts about meetings,\n8:20 notes, that kind of thing. And then just\n8:22 basically build out a system prompt. And\n8:24 then you can pick the model that you\n8:25 want to use to build that out. Now,\n8:26 interestingly here, you can actually\n8:28 then, if you just go ahead and update\n8:29 this, and give this prompt over to\n8:31 Hermes, it will be fully up to date. You\n8:33 can also, within Hermes, if you want to,\n8:35 just effectively ask it build out that\n8:36 skill. I just like to do this personally\n8:38 cuz I like to visualize and see all the\n8:40 skills, so I know exactly what I want to\n8:42 improve and how I want to physically go\n8:43 about it. So, for example, if I just\n8:45 click a feature, I can come down and\n8:46 test this. Either basically go to my LLM\n8:49 wiki and just tell me, for example,\n8:51 three tips on how to make a better\n8:53 intro. Okay, so we can do this, give\n8:55 that to Hermes. And now what Hermes is\n8:57 doing, crucially, is consulting this\n9:00 ever-growing, self-referential,\n9:02 self-improving database of knowledge.\n9:04 Here you go, you come down here, and\n9:05 it's basically consulted this, and I can\n9:07 say, \"Hey, can I just confirm where did\n9:09 you get this information from?\" And as\n9:10 you can see, I already can see here, it\n9:12 read the file exactly where my Obsidian\n9:14 wiki is, and it found the information\n9:16 from there. And look, it's gone ahead\n9:17 and grabbed it. So, effectively, what\n9:19 we've done now is we've got our own\n9:21 individual LLM wiki that contains\n9:24 everything about all the calls that\n9:25 we've had, maybe it contains things\n9:27 about knowledge from experts, and then\n9:28 we have Hermes' own internal memory\n9:30 system, which is by itself incredible,\n9:32 and we combine these two things. Now, to\n9:34 build out your personal LLM wiki, when\n9:36 would you use that? Well, let me give\n9:38 you an example. Let's say for example,\n9:39 you are searching internet and you come\n9:41 across a website like this. This is high\n9:43 agency in 30 minutes, which is a\n9:45 beautiful article by the way, highly\n9:46 recommend it. What you can do now is\n9:48 basically have this index. So, let's say\n9:50 for example that you really like the\n9:51 principles of high agency and what it\n9:53 means to be high agency and how much\n9:54 very valuable. And you kind of want\n9:56 Hermes to understand this stuff when\n9:58 you're talking to it. Like maybe you\n10:00 want to grow on LinkedIn and therefore\n10:02 you want a LinkedIn master article to be\n10:04 in Hermes' memory. Well, what we can do\n10:06 really now is we can copy the text or\n10:07 copy the article. We can go straight\n10:09 over to Hermes. Essentially, what we can\n10:10 do guys is drop it in and be like, \"Hey\n10:12 there, I want you to go ahead and\n10:13 basically index this article into my\n10:15 Obsidian wiki.\" Which means then that it\n10:17 will reference this and understand this\n10:19 whenever we physically want to go ahead\n10:21 and use it. And look, it will just gone\n10:22 down, it's checked it out. This is\n10:24 George Mack's high agency in 30 minutes.\n10:26 Quick tell the offer on the page. Before\n10:27 I create the source page, what do you\n10:29 want emphasized? So, I'm going to say,\n10:31 \"What I'd like to do, bro, is basically\n10:32 embed it into the wiki in line with the\n10:34 principles set out in that claw.md and\n10:37 just kind of index it as part of that.\n10:39 And then from that, it's actually gone\n10:40 ahead and completely indexed it. So, I\n10:42 can now ask it questions about high\n10:44 agency and I'll do that. For example,\n10:46 what are the criteria of a high agency\n10:47 person? Okay, I'm going to send this one\n10:49 off and it will now reference that\n10:51 Obsidian wiki to pull that information.\n10:52 As you can see, high agency people are\n10:54 spotted by four main signals, where\n10:55 teenage hobbies, treadmill energy,\n10:56 unpredictable opinions, blah blah blah.\n10:58 So, the idea here is if you're just\n10:59 talking to Hermes agent, it will\n11:01 remember stuff. Remember, Hermes' memory\n11:03 is incredible. You can even ask\n11:05 questions, \"What did I talk to you about\n11:06 on the 15th of April?\" And it will be\n11:08 able to pull those specific things. The\n11:10 difference here with the wiki is this is\n11:12 an very large knowledge base where we're\n11:14 indexing it. It's a growing corpus of\n11:16 knowledge of things that you find\n11:17 interesting from experts which Hermes\n11:19 can now reach over, grab, and access.\n11:21 And the cool thing is if you're ever\n11:22 working in cloud code, has access to the\n11:24 same knowledge. So, I can say, \"Hey\n11:26 there, do me a favor. What are the\n11:28 spokes of a high agency person? This\n11:29 will will visible in my Obsidian wiki.\"\n11:31 And as you can see, it's on the scale\n11:33 Obsidian ask. It comes on its founder,\n11:35 which is awesome. And look, it's giving\n11:37 us all the information that spoke to me\n11:38 I'm thinking about blah blah blah. So,\n11:40 now essentially Claude and Hermes' agent\n11:42 have access to the same memory. And this\n11:44 is part of the idea of the operating\n11:45 system that we universalize this\n11:47 knowledge and everything's connected in\n11:49 one system rather than having a thousand\n11:50 different things and a thousand\n11:51 different interfaces to use. And so, one\n11:53 of the other cool things that we can do\n11:54 here and if you think about what we've\n11:55 done here, we've wired the memory both\n11:57 ways. So, we can access this LLM Wiki\n12:00 and we can auto ingest files and data\n12:03 into this Wiki without polluting the\n12:05 memory of Hermes' agent, which I'm going\n12:07 to show you in a second. It's very, very\n12:09 cool. And we have this beautiful two-way\n12:10 system giving Hermes this kind of access\n12:12 to this whole new external system,\n12:15 external knowledge. And so, if you think\n12:16 about it, we've got this great\n12:17 bidirectional relationship. Hermes knows\n12:20 everything about you and then we have\n12:21 this LLM Wiki where Hermes itself can\n12:24 just send things over there. Like, if\n12:26 you have a really big conversation and\n12:27 you really want to crystallize that\n12:29 knowledge, you can essentially just say,\n12:31 \"Hey, index this into my Wiki.\" So, it\n12:32 grows this growing corpus of knowledge.\n12:35 Equally, anything from calendars, Gmail,\n12:37 meetings, we can just set up auto\n12:39 basically automatic cron jobs, also\n12:41 known as just like things that run in\n12:43 the background to save that. And we can\n12:44 do that all within Hermes' agent. For\n12:46 example, I could say, \"Hey, what I'd\n12:48 like like you to do is set up a\n12:51 recurring task on every day that goes\n12:53 through to see if I had any new meetings\n12:56 that are shown in Granola. And if I\n12:59 have, I'd like it to add those, please,\n13:01 to my Wiki under kind of like meeting\n13:03 notes, meeting information. Okay? Go\n13:06 ahead and save that one off above, which\n13:07 is really cool. Now, Granola is just the\n13:09 app that I'm using to basically record\n13:11 meetings that I go into. They're not\n13:12 like a sponsor of the video or anything\n13:14 like that, but they're really cool cuz\n13:15 it just basically lets you chat to all\n13:17 of your meetings. And now I'm going to\n13:19 have this database, all this knowledge I\n13:20 can just reference whenever I need to.\n13:22 And there you go, asking us a question.\n13:23 So, I could say, \"Hey, why don't you go\n13:24 ahead and just run it at 9:00 a.m. every\n13:26 morning?\" And I'll say, \"That sounds\n13:27 like a great place to store it. Just add\n13:29 it wherever you think would be most\n13:30 appropriate inside that database.\" And\n13:32 just like that, it's now created a job.\n13:33 So, essentially, every single day now,\n13:35 that will run, it will check all my\n13:36 meetings, if I've got any meetings, it's\n13:38 now added to my LLM Wiki, which is Cap's\n13:40 whole memory system in a nutshell. And\n13:42 then we have access to that in Hermes\n13:44 whenever we want to. And the coolest\n13:45 thing about this, guys, is you could do\n13:47 this with Notebooks. Like, you can\n13:48 connect Notebook LLM to Hermes Agent.\n13:50 I'll put a video on screen if you\n13:51 haven't seen that. Or Cloud Code, get\n13:54 like, you know, 50 to 100 different\n13:56 areas of like expert knowledge from\n13:57 Google's number one research and\n13:59 intelligence platform, I can ingest it\n14:02 in my local Wiki on my computer, and\n14:04 then Hermes can just reference that\n14:06 whenever I want to with this skill. And\n14:07 Hermes itself can create automations to\n14:09 ingest it, so your knowledge base just\n14:11 grows exponentially and will just never\n14:13 forget the stuff that it needs to know.\n14:15 Now, memory is great, but it's only one\n14:17 part of the puzzle. You need to\n14:19 understand how to leverage all of the\n14:21 different aspects of Hermes if you want\n14:22 to get the full capabilities. So, the\n14:24 next thing I'm going to do is learn\n14:25 these capabilities by watching this\n14:27 video right here."
}