Give Me 20 Minutes and I'll Show You How To Use Hermes To Scale Your Company

In this 20‑minute video the presenter explains why most teams under‑utilize the Hermes AI agent by treating it like a simple chatbot and shows how to transform it into a strategic, multiplayer "single brain" that drives company‑wide leverage. The core framework is turning raw context into reusable artifacts (PDFs, HTML, checklists, diagrams) and then converting repeatable actions into "skills" that can be shared, upgraded, and executed across the organization. The speaker demonstrates Hermes’ flexible model switching, integration with data sources (CRM, analytics, Gong, HubSpot, Obsidian, Linear, etc.), memory systems, and permissioning. Multiplayer workflows are illustrated via Slack, with examples of CRO testing, content creation, and sales lead queries. The talk covers the importance of cron‑jobs for recurring checks, the need for governance around read/draft/write access, and the advantages of using the Hermes desktop app for persistent context. Finally, an eight‑level adoption framework is presented, guiding users from basic Q&A to a fully governed company brain, and the presenter promotes SingleBrain.com as a managed revenue‑agent service that builds on these principles.
Chapters
Most people use the Hermes agent like a regular chatbot, meaning they use it like a ChatGPT. And if you're doing that, you're missing out on 90% of the leverage. If you use it as a personal productivity tool, you're not gaining the most from it. So I spent hundreds of hours using Hermes, and I'm here to tell you how you can use it not just for yourself, but for work so you can grow a lot faster. I'm talking about using it in a multiplayer sense, how you can best use it from a workflow standpoint. I'm talking about how you can best use it to skillify things and a whole host of other things that will make your life a lot easier. Let's get into it. So as we get into it, I want to give you a framework here. So the model is context, whatever context that you have will become an artifact. An artifact could be like a PDF or an HTML file that makes it easy for your team or people working with you to be able to digest it quickly. And also your agents can digest it quickly so we can move a lot faster. Now what ends up happening is as these actions become repeatable, they become skills. And what's cool is your team can take those skills, they can upgrade them, or other agents could just take those skills. They can take them and do as you please. So the number one biggest mistake is people use Hermes like a chatbot. And that is such a big mistake because my Hermes has a lot of context. It's connected to a lot of different things. It's connected to a lot of different documents. It has a memory file as well, which we'll talk about in a second. You can just look at how I'm working over here. I'm like, hey, if you look at the X API analytics, last month we got about 3.1 million views. This month we're on track for 1.6 million, and it looks like we need a banger to win. So I'm trying to figure out how we can get back on track with our X profile. It did this analysis just a little bit ago, and it's coming up with some banger candidates over here. Okay, so this to me, this is a strategic partner because I can think with it. And then by the way, Hermes is flexible, it's open source, right? And so what happens is if I want to switch it to a more powerful model for thinking and planning, let's say I want to switch over to Fable, or let's say I want to switch over to the latest version of OpenAI, then I can— that's— it's no problem. I can create another profile within Hermes to do that. So one of the things that you want to do is you're going to want to create a separate profile. But separate profile, not just for duplicating, but you want it for different levels of different tasks, different levels of thinking. And that's going to help you strategize here and come up with better strategies. And so here's a good idea on the, on the, on the right side over here. So it says your website should be a revenue agent. Okay, that's kind of contrarian, because most people aren't thinking about that. By the way, if I can show that we've made our website into a revenue agent, and it's driving more leads for us, that's really interesting, right? Because right now, I'm actually running a CRO test right now. That's a loop. It's the baseline conversion rate on the website is 1.80%. So I said, hey, the target is 30% lift over the next 4 weeks, and here's the target. And it's also going to say, oh, it even comes up with a resource call to action over here where people can comment for it on, let's say, LinkedIn or Instagram or X or YouTube, whatever it is exactly. And so that, that's that, right? So this is pretty helpful because this to me is a strategic partner. I'm not just asking, hey, how should I plan my trip for Italy, for example, which is a very real thing. And by the way, I use Hermes for that too, but that's using it just like you would use ChatGPT. It's just like, to me, it's a more This to me is a more built-out version of ChatGPT, right? But again, you want to use it like, for example, I'm asking it to upgrade scaffolding. So one of my acquaintances, I would say Matt Van Horn, he creates a lot of these cool skills. And so what I often do is I'll just take an X post and I'll just share it with my Hermes. I'll say, hey, upgrade this for us. So upgrade the infrastructure. I'll say, hey, how high leverage is this for us on a scale of 1 to 100? So I'm constantly asking it to push back. I'm constantly asking it to rate the things that I'm sharing with it. That way, to me, it's a strategic thought partner, but it also can execute for me as well. So then you can see here, 34 replies. It looks like it did upgrade— it upgraded the scaffolding here. And so it renamed the stale copies. Okay, great. Over here. So I now have the update for this. Not only that, there's also a /learn command that you can use inside of Hermes that goes a long way. And over here, by the way, the impetus for this video— I was looking at one of these videos over here, 100 hours of Hermes agent lessons. I was like, okay, this is good for personal productivity, but I don't see a lot of talk around this guy using it for— we're talking about scaling a business, right? We're talking about scaling revenues, and that's okay. I'm like, okay, well, that's probably where my edge is. And then this is where I workshop this video, basically, right? So the basis of this is if you're using Hermes as a standard chatbot, you might be getting 5 or 10% use of it, and it's not, it's not a good use of your time. So what I would say is if you want to get even more use, then you should think about how the artifacts will help scale your thinking across the board. Now, what do I mean by that? An artifact might be checklists, it might be carousels and things like that. So for example, this video that I made, I actually asked it to spawn short-form videos and asked it— so it spawned the outline for this video, it also spawned a short form too. So after I do this recording, I'm probably going to record these short forms. And you can see that there's a bunch of ideas here. Okay, so stop using Hermes like ChatGPT. Guess what? I just led off with that, right? But I can create— I can craft this as a short form as well. There's also a checklist over here that people can, people can download. And I don't like the look of this one, I'm going to ask it to, to redesign this. But the whole idea here is that people can comment this and they can get this inside of ManyChat, which is known— this is nothing new from what you've seen. And the cool thing is you can actually watch me do this right now. And so I can just say, you know, AI Work OS checklist PDF needs to be using the ClickFlow design MD for it. Try again. So we have, we have skills in here, which we'll talk about in a moment. We have, you have MD files in here. It has good memory. So we, we actually plug all of our memory. We have QMD, which is a co-founder of Shopify. That's his memory system that we have plugged in here. We have GBrain and GStack from Gary Tan. What we can invoke that whenever we need to do deeper reviews and things like that. And we also plug our daily logs and our decisions in with Obsidian. So there's a lot of context that's being saved and Hermes does a good job of pulling more from, from recency. So that's all I'd say, but look, it's, it's, it's working right now. Like it's making these artifacts. If I like this process right here, I can, I can make it into a skill. In fact, I'm going to, I'm going to queue over here and say, where are we at with making this process into a skill? Right. Because again, a skill is something that you want to, you want to be able to make it portable. You want to hand it off to other people. You want to be able to use it in different harnesses. Do you want to use it? Maybe you can use it in Hermes right now. You want to bring it over to Claude? You want to bring it over to OpenAI? You can do all that. That is the power of this. But I would just say that I'm always making artifacts now. I'm always sharing artifacts with my team. So it might be like a diagram like this. I'm like, hey guys, like here's a diagram. Here's how to think about— here's how to think about this concept. Because people understand if you're sharing like one key idea that's diagrammed out with them, they get it a lot faster. And oftentimes they'll be asking about, hey, like how does that workflow look? Oh, can you share that with me? So a picture is worth 1,000 words, right? And so that's why I think artifacts are heavily underrated. And when you make them into— they could be HTML files, they could be PNG files, they can be— these are downloadable resources over here. But this is me just working back and forth. And it all started with me at lunch watching that video and saying, okay, that's a good title, but what's our edge on this? And how can we do better? And then it literally interviews me on it. And in fact, you know, I'm like, I'm just like, how do we just, you know, beat that video on our own terms, right? And it's like, oh, Here's the different titles you can choose. Here's the different— and then it's like, oh, it knows all the genuinely differentiated ways that we work together. So we're talking about these operating artifacts, we're talking about these different review surfaces, taste filter, right? We talk about these different skills that I'm using all the time, such as Grill Me, Content Machine. There's all these things that I have, and I feel like I'm constantly upgrading my, my weapons every day, my weapons and my armor. That's what I feel like I'm upgrading every single day. So artifacts are very important, especially if you want to scale your communication. Because at the end of the day, if you're just one person, there's only so much you can do. So if a Hermes can give you a private answer, that's useful. Now, if it gives your team an artifact, now we're talking about leverage. Now, the other piece of using Hermes the right way when it comes to work is making it multiplayer. Now, a lot of people right now are talking about these company brains. Let's not overcomplicate it. This is a version. This is an easy way to think about company brains over here if you're using a harness. So in this case, it's, it's Hermes, which is your driver. Then you have different connectors and then you have memory connected in and you decide on the level of permissioning you want to give to your team. And so for us, we have an API gateway with all the different products that we have. That way it prevents our team from kind of just YOLOing and doing whatever they want. And then we tied in with Slack, which is our surface of working. You can see here it's multiplayer. So you can see Chris is asking about, hey, what are the open sales leads for the last past 4 weeks, including dates and follow-ups? And it gives a bunch of information. Here I ask, hey, Who are all the people that submitted their applications in the last 7 days for our Beat Claude challenge? How did they do? A lot of engineers that are applying. By the way, we're hiring a bunch of engineers right now. We're hiring a bunch of AI automation interns. There's a bunch of people we're hiring right now. If you go to singlebrain.com/apply, you can learn more there. All that to say is when you have multiple people in here that are asking questions and they're learning from each other. So for example, if I watch Chris here ask these questions and I look at what the responses are in there, that might spawn another question from me, or that might spawn a conversation that I might want to have with Chris or somebody else. That might spawn another idea where I just put another quick meeting together and then we talk about the actions that we're going to take. But the whole idea here is that the faster that we trade ideas, the faster we're able to grow as a company, because ideas ultimately lead to more action, ultimately lead to better results. That's how you should think about it. So a lot of people like to say, you know, they're using Telegram. Telegram's fine. They like to say they're using iMessage for it. Some people say Discord. But now Hermes in Slack becomes a multiplayer game. So the other problem is if you're just using things like ChatGPT, you like prompting, but prompts are disposable. You want to have skills and eventually you want to transition all the way up to loops. Right now, for the context of this video, we're just going to talk about skills. And so skills compound, like I was saying. So we have a Grill Me skill that I use frequently. And what does the Grill Me skill actually do? Well, it's from this guy, Matt Pocock, and basically it'll ask me questions one at a time, one at a time until it feels like it is satisfied and then it'll create a better plan. The beauty of that is most of the time when you're running through like a planning session, It's going to ask you a bunch of questions at once. And but the problem with that is like, well, what if question 1 affects question 2 affects question 3? So you can't just ask 4 questions at once. And so I like Grill Me because it does push you and it'll keep pushing you until it feels like it's satisfied. So sometimes it might be 1 question, sometimes it might be 12 questions. And someone like me who's not a detail-oriented person, it's annoying, but I like it because it forces me to think about what we're trying to do a little more. And usually I get better results that way. So garbage in, garbage out, right? But if you spend the time on it, it's going to be a lot better. I have this Content Machine interview too. And so, by the way, I'll just show you the Content Machine. So give me a second. Okay, so you can see here what's happening as I'm working through this is I'm just yapping through it. I'm like, hey, run the Content Machine interview to get my different POVs. And I'm like, hey, you know, he said all these things. Some things are cool, some things I believe are useless. It says, hey, here's my read before interview. Your strongest POVs are hiding in the dump. Kanban is the wrong hero. Hermes for work is multiplayer. The artifact is unit of leverage. So you see what's happening over here. This video becomes a little better because I spend time, I spend the actual time to plan it. And I'm curious, let me, let me know below if you, if you do agree with that or if If you don't agree with that, right? So there's a lot to— that's going on here already. There's actually a lot of meat already. And then it starts to grill me over here. So what's the single best real work artifact Hermes has created for you or your team so far? Could be a brief, teardown, prototype, memo. Honestly, it's creating the Skills Dojo, right? But because I felt like I had enough already, I just moved on to the next thing. And then I was like, hey, what other generally differentiated ways do you see us working together that should be folded into this video? Because you know why I can ask that question? Because my Hermes knows how I work. It has memory and I don't need to repeat myself over and over. Okay. So then it's like, okay, here's the generally differentiated ways in how we work. But again, you see, I'm not using it like a standard chatbot. I'm using it in a very different way, like a very highly paid thought assistant. People like to call it chief of staff. I think it's more than that, but that's what it is. By the way, if you're enjoying this video on compounding Hermes into a single brain, well, that's what we do. We have Single Brain, which is our managed revenue agents that we build for our clients. that help compound these systems, right? Whether it's stalled deals that need reviving, or we are talking about building complete outbound systems from scratch with these agents, or if we're talking about scaling creative, or if we're talking about an AEO SEO agent that can compound the SEO AEO work that you're doing, it is all there. Just go to singlebrain.com. That's with a B, and we'll see you on the other side. So the way I look at prompts are prompts are, they're hard to remember ultimately, and they're very fleeting. But when you think about skills, skills are infrastructure for you and skills make it fun. Like when you have a Skills Dojo like what we have, and by the way, if you go to skillsdojo.com, you can actually get access to our public dojo of skills that you can take, you can fork, you can download, and we want to make things available to everyone so we can have more fun for people to get access to these skills, right? Because you all have a lot of interesting things to share. So skillsdojo.com. And here's one of our skills right here. Here's my ex-long-form post writer. So this is great because what happens with this X long-form post writer is I created this a few months ago, and then what happened is that, you know, people can upvote it, you can bookmark it, you can fork it. But the whole idea of these skills is that it's basically a process. You can— and I don't think people use skills enough even today. You have to have a good skill. There has to be a good eval to define what the definition of done is. But this, this basically gives a formula on how to write an X post. And then over time, if other people start to use these skills, They can fork your version, they can make better versions of it that end up getting more upvotes. But that's why we think it's important to be creating more and more skills. Like anything that you repeat more than once, you should probably make it into a skill. And when you use Hermes, I just ask it to skillify the things that I have. And I think that's inside of G-Stack or G-Brain. But the whole idea is like, hey, like I just want to make this into a skill. So I'm constantly thinking about what should I make into a skill? What should I make into a loop, right? How can I just add more leverage every single day? So I'm constantly asking myself, How high leverage is this 1 through 100? And then I asked Hermes, how high leverage is this 1 through 100? One of the most powerful questions that I like to ask. And then we have a dojo section over here for these skills, right? So you can pack up these skills for different roles. So that's what I would say. So by the way, this Skills Dojo, the original repo that was created by me using Hermes. So you can make anything that you want, almost anything that you want. And you can use whatever models that you want. And I think that's, that's a very beautiful thing. You can use it on local models. You can use it on your subscription. You can use it on our API. I think just having that flexibility is extremely powerful. One more thing I'll say is that the best AI agent is not necessarily obedient. The best AI agent is a taste shaper. So what do I mean by that? The Grill Me questions forces deeper thinking. When I'm using the Content Machine interview, it forces the best content out of me in that moment. And also I use Hermes to cut out a lot too. There's a lot that needs to be deleted. And then I look at what we have with our YouTube packaging skills for long form or for short form. I look at what we're using for recruiting skills as as well. I look at what we're using for AEO SEO skills as well. So ultimately, Hermes will tell you if it thinks your idea is mid. It'll also tell you if it thinks you're not doing a good job. It'll also, by the way, you can call on any of your connectors whenever you want. So what do I do here? I will ask Hermes, I'll say, hey, look at our HubSpot, look at our Gong. Tell me what are the patterns that you're seeing from our client calls that we're having? What are the patterns that you're seeing from our discovery calls around SingleBrain, which is where we sell these managed revenue agents? What are the patterns that you're seeing with single brain? One of the questions I asked recently was like, hey, on a scale of 1 to 10, how excited are people to talk about single brain versus single grain? And it literally said 8 or 9 out of 10 when it comes to single brain. When it comes to single grain as a service, not as high, right? And so to be able to do these data pulls a lot faster, that allows you again, the faster exchange of ideas allows you to move faster as a team because you make decisions a lot faster. So my next point is around connectors. And the magic around connectors aren't the connectors themselves. It's around the synthesis of information with these these connectors. Because like I just shared with you right now, if you have all these different things like analytics, your CRM, your revenue intelligence, all these things talking to each other, and then it's also talking to your Google Docs as well, what ends up happening is you end up making better decisions, right? And then when you run it with a, with a more strategic model like a Fable or like the latest version of OpenAI, you're going to get better outputs, and that's gonna, that's gonna, that's gonna better inform your builds. Because sometimes when you have information that's slightly off a little bit, that can bring you in the wrong direction. Because remember, when the airplane's off by like 1 degree or so, it ends up in a completely different part of the world. Look, there's a lot of different things that you can attack here, but you can't— none of it matters if you keep using this like ChatGPT. What matters is you are building leverage with company intelligence. So I'm going to give you a framework here when it comes to thinking about how you build this. Okay, so when it comes to read, okay, people can read documents, that's probably okay. When it comes to drafting, you need a little more scrutiny there. When it comes to writing, which means executing and publishing stuff, you need way more scrutiny. So read access is usually fine if you scope it properly. Drafting is where Hermeus should be aggressive. And then writing externally should require human approval. That's why it's very important for you to think about this sooner than later, because if one thing goes wrong with a client or one of your employees, for example, and let's say all your information gets leaked to the web, well, that's ultimately on you because you didn't think about this. But if you think about this, you think about governance, you think about permissioning, you think about security and all these things upfront, you're going to end up doing a lot better. So one more thing I'll say is that you want to make sure that you're using the Hermes desktop app. The Hermes desktop app is very powerful because the way I use it inside of Slack right now, it's not as good because Slack is very fleeting. If you want to use Hermes correctly, you have to use the Hermes desktop app. The reasoning for this is because you can actually see what you were working on before. You could actually ask it to reprioritize on the left side over here. You can look at the skills and tools that you have. You can look at the different artifacts that you've created. You can look at, you can look at cron jobs that you've made as well, which we'll talk about in a second. But the whole idea here is that it's easy for me to remember what I'm working on. Versus when I'm working inside of Slack or Telegram or iMessage, it's very fleeting and I'm going to forget. And each of these apps has limitations, but this Hermes app is going to only get better over time and it's already really useful. And so it reminds me of working inside of Codex and Claude Code, which I still use all of them. And the reason I like using those is because it easily reminds me on what I need to get back to. And the cool thing is whether you're working on Telegram or you're working inside of Telegram, you're working inside of iMessage and you switch over to Slack, and then like all that context, you can pull it in over here from all these chats that you're having. So I think it's very important that you're using the Hermes desktop app. Related to that are your cron jobs. So your cron jobs are these chronological jobs that fire. Maybe you need a reminder, hey, here's how your pipeline is moving today. Here are the emails that you need to follow up with. And so you need to be setting up these cron jobs, thinking about deals that are stalled, follow-ups that need to happen, right? And so a good cron job is something that tells you, hey, here's what's changed, and then here's the action that I'm recommending. And then also, would you like me to do it for you? Just be very careful that you don't overdo it with the cron jobs, because a lot of cron jobs become irrelevant over time. So you probably want to do a scan every now and then to make sure that you have the right cron jobs in place and that they're still relevant and they're still high leverage for your business. So a little more on memory, because I talked about it earlier. If an agent can't remember its decisions, then it cannot become a brain. And we're talking about this whole concept of single brain, right? This, this company brain. And so you don't want the work to evaporate. And so what we do is we use Obsidian for daily logging. We have QMD. There's a decision history. You can see active tasks as well. We also will store things into Linear as well if it's a larger thing. And we have skills that we create from repeatable workflows we've put into a skills dojo. So that's something that we can remember. There's a lessons.md file that keeps getting added to over time. And we have these active cron jobs too, right? And so you want to make sure that what you're building is not just like a ChatGPT because ChatGPT doesn't compound as well. You want to build a durable system that compounds. So 2 more housekeeping things I'll give you. You know, when I'm walking, for example, I'm gonna go on a walk right after this, but you know, when I'm walking, I am also— I'm oftentimes just yapping using WhisperFlow on my phone. I just hit a button over here and then I just talk for about 5 minutes or so on whatever's on my mind. And then I will let my Hermes listen and I'll probably post it into Slack or I'll— you can post into Telegram, whatever it is exactly. And I find that because WhisperFlow does a good job with transcriptions, I can get my ideas out there and then it does a good job of distilling my ideas and then figuring out where we should go next. Maybe there's a skill that we have already, maybe there's something that we need to to build on top of. Maybe there's redundant work that's happening because sometimes I get lost. So that goes a long way, you know, talking about product ideas or YouTube ideas, content ideas, team issues, talk about sales insights, just random thoughts overall. I think being able to, to go through voice is powerful. Hermes has the ability for you to connect from, from a voice angle as well if you want to do that. I just find that I like going back and forth with it. So I'm going to give you the 8 levels of Hermes. Level 1 is you just simply ask Hermes questions. Level 2 is you delegate one-off tasks. Level 3 is you create shareable artifacts. Level 4 is you turn repeatable work Into skills. Level 5 is you turn recurring checks into crons. Level 6 is you connect your company data sources. Just be careful. Level 7, you want to add permissioning and governance. Level 8, Hermes finally starts to become that first version of that company brain that you can actually start to compound. So reality is most people are stuck at level 1. This video hopefully starts to get you towards level 8. That's where I want you to be because I want you to build reusable systems that compound. So I hope you enjoyed this video and we'll catch you in the next one.










