# This OpenSource Repo will 10X Your Hermes Agent

Channel: Jack Roberts
Video: https://www.youtube.com/watch?v=yOZVYw9FIWc
Duration: 15 min
Language: English
Words: 3598
Transcript page: https://viewrankai.com/tools/youtube-transcript/yOZVYw9FIWc

---

[0:00] This new open source AI will 10X your Hermes agent. Now, most people use Hermes agent like a chatbot. They just ask it questions. But, research operating systems are used by companies like OpenAI and Anthropic. And if you apply them correctly, you can unlock capabilities that 99% of users don't even know exist. And so, in this video, I'll show you exactly how to get it set up easily, giving your Hermes agent unlimited knowledge and making it way more powerful so you can stop wasting time, make more money, and get light-years ahead of everybody else. And if you're new, I'm Jack, a button-smashing startup with a gazillion customers. Now, I'm building my own AI

[0:37] startups, and I just share here the stuff that actually works. So, if you haven't already, grab that beautiful coffee, and let's dive straight in. Now, when we're thinking about our beautiful Hermes agent, the best way to think about the way that most people use Hermes agent is a genius locked out of a library. When you ask Hermes agent to do something, it can do a very basic internet search. But, effectively, there's some stuff it just can't grab, right? grab transcripts from YouTube. It doesn't have access to every page on LinkedIn or Reddit or X. Therefore, it's not getting the very best knowledge available on the internet. But, if we

[1:11] could in theory connect it, I think you could have access to essentially everything that's happening right now in the world. Now, in this video, I'm going to show you two-level upgrades, okay? One on connections and one as a specialist that's going to drastically increase the performance and the knowledge and the research capabilities of your Hermes agent for anything that you want to go ahead and do. So, the idea here with this new brand new GitHub repo, actually, is that we can give our Hermes agent eyes. That's the whole idea, and the repo is something called

[1:39] Agent Reach. It's got over 32,000 stars. It's number one trending right now. It is blowing up. And the whole idea of this GitHub repo, and I'll show you right here so you can have a look. It's this one right here that we're hanging out with. I'm going to dive into that with you in a second. If you do make it to the end of the video, which I highly recommend, you are going to have superpowers in your Hermes agent. People will be saying this isn't fair. How the

[1:59] hell is your Hermes agent so powerful? So, what can we do with this repo? Well, we can give it access to X, LinkedIn, um we can basically RSS feeds, GitHub, anything that you possibly want to. This OS is even updating itself in the background. That's how much cool stuff's going on with it. And it basically the idea is that you get access to lots of different softwares, and it has different fallbacks to make the search more robust for sure. It works way better. And not only that, it's going to

[2:24] give us clean data and better answers. So, it returns structured text, not raw HTML. So, the same research costs a fraction of tokens. So, it's literally going to save you money the more that you're actually using it. This is an example. You could have like 86,000 tokens for the HTML, in other words, the website speak. But with the clean structured extraction, you know, you can

[2:43] get 30-40% saving you loads more time. Again, 2.5x saving on the tokens, and that's really important to understand, guys, because the Hermes agent itself has a specific context window. And one of the biggest complaints, this actually solves that. That's why it's so, so important. And the good news is it's super duper easy to install. So, we're going to go ahead and install this one together. So, what we're going to do is head over to this GitHub repo right here, and I'm going to go ahead and grab the code. And all we're going to do is

[3:07] come down here and copy this. Beautiful. So, we'll light up the operating system, and we're going to have a conversation with this voice intelligence lab, which is going to make it super cool. Let's come down and begin a conversation with Hermes agent. Okay, come down and we can start off. Hey Jack, what's on your mind? Hey there. Basically, I've got a GitHub repo. I'm going to give it to you. What I'd like to do is go ahead and install that for me, uh so we can use this new

[3:28] research engine. Sure, I can help with that. Let's get the details of the repo, and we'll go from there. What's the link to the repository? Cool. I'm just going to enter it into the chat right now, and then let Hermes work its magic in the background. Let's start by cloning that repo and setting it up. Are you planning to run this on your local machine or on a

[3:45] server? Yeah, let's let's run that on the local machine, please. All right, running it locally it is. Let me pull in the information and get the installation process going. Beautiful. So, Hermes has done some work in the background. What I would love to know as well is could you just make a list of all the new capabilities that this repo has given to you as Hermes agent? And then let's think of an example to test this just to make sure

[4:06] that it's completely live right now. Hermes now has the following new capabilities from that repo. Advanced web scraping, real-time data extraction, fine-tuned natural language processing. Okay. So, you get the idea. Basically, Hermes gives that full breakdown. Of course, if we prefer, we can just have a regular conversation with it if you want to. But it has access to all these different functionalities now, which is great. Of course, I've also got Telegram up. So, if you prefer you just type it on the go, we can do. Cool. Hey, I just downloaded the brand new skill from GitHub. What I'd like to do is to wire this into Hermes agent so that you can

[4:38] use it yourself. And then just give me a short short overview of the new skills this has given you and how this improves on what you had previously. And then we're going to test this live together to see that in action. And I like to do this as well, mainly because it's great for Hermes to explain and articulate exactly how it works. And I always do

[4:55] this whenever I get a GitHub repo. Hermes has read and just sort of understood everything anyway. So, it's good to give you that breakdown back in terms of what's working and how you can improve on it. So, Hermes come back. We've got new useful routes. And this is a good way of thinking about it. It's not making Hermes more intelligent. What it's doing is improving its routing, its intelligent routing system. So, Hermes doesn't need to rethink every time it wants to do something. It knows it's got the exact processes that have been outlined in the skill, which is fantastic. And again, it knows which tool to use for which platform. It can

[5:24] run a live doc to check out and prove what is working. And it gets better fallbacks when normal web tools fail, which is really freaking handy. So, instead of summarize this repo, you say let's grab this repo and see if it's worth making a video about. And then Hermes can do his magic. So, what we're going to do is give it this little prompt here, which is go ahead and wire everything up. And the beautiful thing is Hermes agent can go ahead and do that for us. Beautiful. So, now Hermes has

[5:45] come back and connected that together. And by the way, if this sounds like I'm speaking Japanese, I'm going to link down below for the Claude Code full course that will take you from foundation setup, website, a full Hermes masterclass, everything you need to know to get up, get started, compliance stuff I've never shared on my channel, as well as this entire Claude Code operating system, the voice visual intelligence layer you just saw me, and all of the bells and whistles. I'll put a link down below so you can go ahead and grab that one. Now, one of the things you want to do here is just confirm first of all

[6:12] what is active and put it to the test. Probably one of the best tests we can do is YouTube. So, we've got web pages, and it's just given it now more intelligent routing. So, when we say to Hermes, "Hey, go find out this information." it's got that system that's already built into it. So, some stuff that needs connecting, and then I'll talk about that in a second. So, and how this actually physically works under the hood, if you want to get X router, all those different bits and pieces. So, I'm going to say, "Awesome, let's put this to the test. I would like you to go over

[6:36] to the Jack Roberts YouTube channel, and find my latest long-form video. And then, why don't you go ahead and tell me what was the first sentence of that I'm using your skill. Okay? Send that one off and let it work its magic." And then, just like that, not only has it found the correct YouTube video, it's given me the exact sentence, which I think is just really

[6:56] cool, cuz now we have access to YouTube. Now, we can do this with GitHub. So, I could say, "Hey there, I'd like to go ahead and test this GitHub. Tell me what this GitHub repo is about, and if it's worth discussing." All right? And I'm going to drop in the actual, ironically, the GitHub that we're talking about right here. And so, we can test this one out. Beautiful. And just like that, it's come back and actually got everything for us. So, give your AI agent eyes across internet, it helps you search for things like YouTube, GitHub, RSS, Reddit, X, etc., etc. Here's a key nuance, it isn't a magical scraper, it's

[7:25] more like a routing and install and doctor system. So, which tool to use, which background's currently working, what needs authorization, what's broken, etc., etc. Now, how this works give you access to things like LinkedIn, Reddit, and X, is it does it based on a cookie system. So, essentially, the way that this is configured is that you sign into the services on your computer, and then it uses that cookie. So, that's fine, so to speak, but it's only really as sort of like balanced and functional unless something changes in the background, it might not work. I personally don't like going down that route. So, what I do

[7:57] instead is leverage other systems I think are way more robust. For example, if you connect within your Hermes agent, you can actually connect Grok very, very easily with the Grok OAuth, and you can do this in terminal also. But, what we can do here, if you actually go ahead and use Grok, you can search the entirety of Twitter. So, for example, if I come off this, I might come back over and say something like, "Hey, dude, I'd like to go ahead and use Grok and find for me like five trending and interesting developments with Claude code in the last week. I want to know ones that have say good view velocity,

[8:29] good engagement velocity, and everything like that." So, you can configure this using Claude or Hermes itself. Just bear in mind that the repo itself is in Chinese, so if you're a medical doctor, you may not understand it unless you basically speak Chinese. I personally, though, I'm not using any of the cookie things cuz I think things like X I can actually access myself within Telegram by connecting to Grok. But, if you did want to bring in those different things like Reddit and you didn't want to give your API key, you can use this kind of strategy to grab the cookies. And then just like that, it's gone ahead and used

[8:57] Grok, and now we have access to the entirety of X all within Hermes agent. How cool is that? And all you have to do is just connect X to it in the model select, which is really cool. Now, this comes back with my ideas of dynamic workflows, things around agent loops, lots of conversations around that that have been around for a very long period of time. What's really cool is actually come down here, and use Grok X for the search tool, and then I actually went ahead and used the agent research and the YouTube like basically YouTube search to sanity check it to make sure that it's fully aligned, which I think

[9:25] is really freaking cool. And it leads us very nicely down onto the second part of this, and that's essentially this idea that we now have this research machine in our pocket. This idea that we can yap to a Hermes agent whenever we want to, and it goes and finds out all of this wonderful information. But, we can take it one step further, and this is where we get to level two. Now, what if I told you that what we're going to be using right here now is actually used by

[9:47] OpenAI themselves, and also Anthropic. So, it takes it to a bit of different level, and just gives different capabilities to our Hermes' agent. And I'm talking about clay.com. So, this is used by a gargantuan different companies, which I thought was super interesting to pique my interest initially. HubSpot, they got case studies with what? Canva, OpenAI. Very, very cool. So, think of clay as like almost like a database or a table that fills itself in. It's like a go-to-market data platform, and it enriches leads from like hundreds of different sources. So, information is one thing, but when we want to grow our business or do various other things,

[10:23] clay is a super duper kind of big system that can work in the background. It's got agents. It's got lots of really cool stuff. We're going to integrate it into Claude, and also Hermes' agent to give it some really powerful capabilities. And when I reached out to the team, I said, "Dude, I want to do a video with you guys." They graciously agreed to sponsor this part of the video. So, thank you, Clay, for doing that. But, let's start actually by logging in. I'll show you exactly what I mean, and why this needs to be on your radar and why it's so cool. So, we're going to come

[10:47] over down here and just sign in with Google real quick. So, first of all, if you wanted to take your signals to another level, one of the things we can do, you click on signals on the left-hand side, and let's say that you and I want to target dentists or you and I want to target, I don't know, doctors, whatever it could be. We can actually start using some of these different categories to do that. So, if lots of stuff that you can do within the actual clay platform itself, but we're going to be using the connector. So, we can actually use it within Hermes directly,

[11:12] and also Claude. You can come down to the plus button down here, get down to connectors, and you can add it as a brand new connector, which of course here is just going to be clay, and you can see the different things you can do. The capabilities are exactly the same within Hermes' agent. Now, I've made it unbelievably easy to connect Hermes with clay. I'll put a link for this down below, so you can literally go ahead and grab it. All you're going to do is come down, going to copy all of this like so, head over to Hermes, and just drop it in. And then once that happens,

[11:36] essentially it's going to open up some of these, authorize yourself, and if not, you can just run this in the terminal. Going to come down here, this is MCP access. So, just pick the workspace that you want to give it to, come down, and click on authorize. And as you can see, authorization is now successful. So, when you pick your workspace, if the first one doesn't work, you can actually check out the second one, and that will work for you absolutely fine. And then to validate that, we can just say to Hermes agent, "Hey there, could you just confirm that you have access to the Clay MCP,

[12:01] please?" And it's cool, cuz then we can turn this into a skill, so we can use it whenever we want to. We'll just get Hermes agent to confirm it, and then we can test out the magic of Clay. And just like that, it's now fully connected. So, let's give it a prompt. "Hey there, so I would like you to, based on everything you know about me, go and grab for me three people from Anthropic, and three people from OpenAI, that I could make contact with. I want you to do some research on them, based on all the things you have available in Clay, and write for me a personalized message for

[12:30] all six of those individuals, who I, as a YouTuber, um founder, entrepreneur, could make contact with, to work with those organizations. And then I'd like to go ahead and create for me a HTML overview, so I can see all the messages, background information, research, but also feel free to publish that in the chat, first of all." So, now and off, pretty tough challenge, but again, if this information is amazing, we will absolutely crush it with this. And just like that, Hermes is completed. We've done everything with Clay, which is fantastic. It's even created a HTML document, we're going to check out in a second. It's got six people, three from

[13:01] OpenAI, and three from Anthropic. And he's explained why that's helpful. So, I'm going to pull up the HTML document that it pulled together, which I thought was really helpful, actually. The fact that we can do this with Hermes, again, is freaking awesome. Six air partnerships for Jack. Let's check out exactly what we're dealing with here. So, best angle is pitch yourself as a creator founder, blah blah blah. And then we've got OpenAI, why him, collaboration idea, message, message, message. So, So only do we have the six contacts, we've also got a figurative message that we can send there. And then one thing I'm noticing here is we don't

[13:30] actually have the contact information because I didn't ask for it. So we're going to go straight back over to Hermes and ask Clay to go ahead grab us those emails as well. I want to come back. It's going ahead. Clay has found four out of the six so far, two are in progress. It's just polling. And it's just so handy to be able to say, "Hey

[13:44] dude, what are these email addresses?" And be able to get those just from a question. Beautiful. And now we've got all the emails back. I'll blur them out for their privacy, but you get the idea of what Clay is possible, especially when we're looking at things like incorporating with Hermes. And this is the gentle caress, this idea that everything we have sits together in one specific location. And of course from that, we can do a gargantuan different things, have those live conversations if you want to, which is cool. So Clay itself, and this is important to understand in terms of what it can do, why people are using it, verified

[14:10] emails, can do AI research, find signals, sync with CRMs, and do individual outreach. Specifically, this is what I mean. So they've got a waterfall across 150 providers, and the pricing only hits when you find a hit. So you're not basically charged for something that doesn't work. We can find committee, so find decision makers, which is cool. They've got a Clay agent AI research, which is quite handy. And then if you want to build up the sequences, you've also got a personalized outreach. But it does lead me on very nicely to kind of the crux of this entire intelligence system. And that's the fact that we're building out

[14:40] this research intelligence agents. So then when we're using Claude and Hermes, whichever platform we're actually going ahead and talking to, we can actually get the right information as quickly as humanly possible, saving us time, saving us failures, and saving us a lot of money because it's not failing. So now we have a research intelligence system that, without a proper operating system behind it, isn't going to get us as far as we want to go. So what we need to do next is build that operating system,

[15:04] which we do in this video right here.

---

About this transcript
Read from YouTube's own caption track and laid out by ViewRank AI (https://viewrankai.com).
ViewRank AI finds the videos already beating a creator's own average on Instagram, TikTok and YouTube Shorts, transcribes them from the audio itself in more than 60 languages, and turns what worked into new ideas and scripts.
Free transcript tools, no account needed: https://viewrankai.com/tools
How to read any video this way: https://viewrankai.com/llms.txt