Claude 5 + Hermes: Build ANYTHING!

**Julian Goldie** (0:00)
So, Claude's most powerful model ever just dropped today. And I want to show you exactly how you can get the most out of it, because most people have no idea how to actually unlock its power. And that's exactly what I'm going to show you today. What we've actually created is five different projects, and even created this video, as you can see right here, down in each of the projects that it created, which is pretty amazing in itself. And even created, for example, like an orchestration dashboard, where Claude can control all of the agents, delegate tasks. We have a group chat right here. We have multiple other things I'm going to show you in a second, where you can get your AI agents to be commanded by Claude, and then also get way more out of Claude. Now, the reason you want to learn this today is because this is only available for a couple of weeks. On 22nd of June, Claude Fable 5, which is MIFOS level as a technology, will not be available on the subscription anymore. So if you're going to use it, start experimenting with it today. And that's exactly what we're going to show you today. Now, if you're not familiar with what Claude Fable is, let me just give you the too long didn't read. So on June the 9th, Anthropic released Claude Fable 5 and announced Claude MIFOS 5, which are brand new fourth model classes above OPUS. Fable 5 and MIFOS 5 are similar models, but Fable 5 has safety guard rails. MIFOS doesn't and is restricted to VED partners. And it's state of the art on nearly every benchmark, scoring 95% on SWE bench verified, which makes it super powerful. It's also included inside all Claude plans that are paid until June the 22nd. But from June the 23rd, it needs usage credit. So you might as well start using it now. And Anthropic have disclosed themselves that over 80% of its own merge code now is written by Claude. So this is a super powerful model that we can use. So let's get straight into this.
The first thing he actually created was this whole site, as you can see here. So it created this full guide explaining what it did, how it worked, what it did, the whole framework, and then included links to each of the projects he created. Now, one of the projects he created as well was basically a website where we asked it to generate a landing page explaining exactly what the agent operating system is.
And you can see that here, it's generated this full website and created it with this nice background. You didn't have to prompt it at all, but it just went off and built this.
And then it broke down the key concept of the whole agent operating system, how it works, what it can do, all the benchmarks, how the system works, wrote all the copy, and it's a pretty nice design.
Now, what we also have inside a system is this pipeline, where we can generate and build anything on the spot that we have an idea for. So if we have an idea, we can just plug it in here, capture it, it's going to plan it, ask us to approve the plan, which you can see here. And then once it's done, it's going to implement it. And once it's implemented, we'll see it in the shipped section here. So this is another website that we've built with Claude directly. Now, on top of that, we gave it one idea, which was an agent command center simulator, like a living mission control screen. So obviously, at this point, we've created a full mission control, but we wanted to bring it to life. We wanted to make it more interactive, and we've actually created that here, as you can see. And so we can see what is delegated, what's being done. So for example, like syncing to Obsidian, delegating tasks to OpenClaw, etc. And then we can actually click into each of our agents inside the system. And we can see what is done, what model is running on, what is current assignment is, and we can see all of this in real time, which is pretty insane itself. Now, the other great thing about all of this that we're doing here is that we can see it inside our workspace. So for example, usually if you're using Claude, like everything gets lost or if you're using the terminal of Claude code, you can't find stuff that you've previously created. We've got a full workspace here where we can save everything that we've created previously. And so all this cool stuff that we've built, we can come back to later, we can review it, we can use it for other projects, we can build on it, we can improve it. And that's because we can see everything in one place, which is awesome. Here's another blog that Claude has created as well, which is pretty amazing itself.