**Julian Goldie** (0:00)
So today, I want to show you a new powerful system that we're using with Hermes Agent to basically automate and build apps in one single click. And the way this works is that we can drop an idea right here, then we can shape it with our plan. From there, we can approve or deny it. So for example, if we look at this plan right here, we can click on approve and build. And then from there, it's going to start building inside the implementation setup. And then we can view everything that we've built on this side. And we actually get a live preview with the idea, the classification, the plan that was created. And we can also start or delete stuff that we don't like. So it's pretty cool as a system. And also it means you can go from idea to implementation really, really fast. As you just saw, we went from idea to implementation in like one or two clicks right there. So this is one of the powerful systems we've built into our agent operating system, which means you have Claude, we have Hermes, we have everything else built inside one place. And the great thing about that is that it's very easy to orchestrate our agents. Now, here's something else that we actually plugged into the system, which is we can actually ask our AI agents in a group team chat.
Okay, give me a great idea for this app that we're building. And then our agents are going to talk among themselves, read our Obsidian vault as well as read information about us to get the context. And that means that they'll come up with personalized ideas based on me, my goals, what I've worked on recently, and all the memories inside my second brain, which then is going to coordinate between Claude, Hermes, Gemini, whichever agency got uploaded into the system. And then from here, once we actually get a good idea, so for example, based on the SEO app that we asked for, it says quick SEO gap idea creation. It's going to plug into our pipeline, and from there we can shape, implement, build it. And the cool thing about this is as well, if any of these ideas are not very good, you can delete them. If you're like, I don't like that, let's delete that, you can, no problem. If you really like the idea, you can actually start it.
If you want to check it first, no problem, because what you can actually do over here is, you can decide if you want to approve or deny this plan. So if we have a look at this one, we can see that we can approve and build it, or we can deny if we don't like it. Then from there, we can actually implement it. And from there, you actually get the shipped and filled section over here so that everything that you've created and built out is ready to go, and you can come back to it whenever you want, which is pretty amazing itself. Now, also, what we can also do is click on the gallery section, and we can see everything that we've built in one single place. So the amazing thing is that we can build this huge collection of tools and ideas that we've created, open them up whenever we want, and some of this stuff is super nice. Check out this website, pretty nice. Here's another one.
And the way that we've actually done this is we plugged Minimax into Hermes Agent so that we get a good balance between speed and intelligence. And also, because it's a flat rate coding plan with Hermes Agent and Minimax, you don't have to worry about tokens, right? I've never run out of tokens with Minimax. So I can build whatever I want. I can create anything that I need to. I can move really, really quickly. I can use Hermes Agent to help me. And then I can build stuff and implement it really quickly. So this is the app builder system. And basically you can drop a sentence into the Agent OS.
So for example, it could be like a habit tracker or meditation timer. Agents classify it, draft a plan, and after one approval from you, they build a genuinely working single page app with every button wired in, real data, everything plugged in right there. All right, so you capture an idea, the agents plan it, you approve it, they build it and then it's shipped. So if you look at this section, all you have to really do is approve it because even for the idea itself, your agents can come up with the idea from the group chat. And the point of this as well is like, if you've looked at, for example, Sakanofugu or Fusion or Hermes mixture of agents, having multiple models work together will pretty much always give you a better level of intelligence than you can currently achieve with the best model out there. And I think that's probably a fundamental rule that may be true forever. So it's a great thing to learn. But number two is that you can use that by having a group chat like this where your agents can talk and they can reply, they can riff off each other. They read your Obsidian vault for context. And then from there, they plug those ideas automatically into your pipeline so that you can build and automate whatever you want based on their best ideas. Now you might be wondering, okay, how does a memory system work? So we have Obsidian, which is a free app. We've built all of our memories into this memory galaxy. It's all visualized, organized, linked together and very nice. Basically, the way that we've organized app is that our agents automatically update it. You can see, for example, two minutes ago, four minutes ago, 35 minutes ago, like our memories have always been plugged into this system. Then from there, our agents can update it but they can also pull from it. So it's a positive feedback loop. It's a loop engine where our agents update and organize the memory galaxy themselves, but they also use that to come up with ideas and build amazing things that are personalized to me, so that we have our agents just running on a loop and there's very minimal human intervention involved, unless you need to. And that's why we do have the human approval section here, because we don't just want to build everything that they create. We want to look at the plan, see what it is, see if it's actually good, and then approve it, if it's good. If it's not good, let's delete it.
6 more minutes of transcript below
Try it now — copy, paste, done:
curl -H "x-api-key: pt_demo" \
https://spoken.md/transcripts/1000651996090
Works with Claude, ChatGPT, Cursor, and any agent that makes HTTP calls.
From $0.10 per transcript. No subscription. Credits never expire.
Using your own key:
curl -H "x-api-key: YOUR_KEY" \
https://spoken.md/transcripts/1000774777651