--- title: 'Fable 5 Leaks, Fugu Ultra Review, and AI Design Tips' source: 'https://youtube.com/watch?v=HEh7paFkC0w' video_id: 'HEh7paFkC0w' date: 2026-07-14 duration_sec: 0 --- # Fable 5 Leaks, Fugu Ultra Review, and AI Design Tips > Source: [Fable 5 Leaks, Fugu Ultra Review, and AI Design Tips](https://youtube.com/watch?v=HEh7paFkC0w) ## Summary The video covers the latest AI news, including leaks suggesting Anthropic's Fable 5 model may return soon, a detailed review of Sakana Fugu Ultra's performance, and a tutorial on improving AI-generated web design using open-source skills. The host also demonstrates his Agent OS system for automating content creation and SEO. ### Key Points - **Fable 5 Leaks** [00:00] — Claude Code V2.1.190 and AWS Bedrock documentation show string changes hinting at Fable 5's return, possibly within 3-5 days. - **Hermes Oracle Demo** [02:00] — The Hermes Oracle pulls trending news and can draft social media posts or publish articles to WordPress automatically. - **Memory System Setup** [05:00] — OMI takes notes, syncs to Obsidian, then feeds into Agent OS to create an automatically updated knowledge graph. - **Sakana Fugu Ultra Review** [08:00] — Tested 42 builds; Fugu Ultra creates impressive games but is one-shot and slow (15-20 min per response). Outperforms Opus 4.8 but trails Fable 5 on some benchmarks. - **Google June 2026 Spam Update** [12:00] — New spam update rolled out June 24; uses SpamBrain AI. Advises waiting before making changes and focusing on helpful content. - **Agent OS Community Showcase** [15:00] — Members share custom setups: Joseph built his own memory galaxy, Sheena added a search feature, Scott automated content with Hermes. - **Design Skills for AI Websites** [18:00] — Using open-source design skills (e.g., shadcn/ui) to avoid generic AI look. Principles: one bold element, font with character, single color, white space. - **GPT-5.5 Instant Release** [21:00] — New model from OpenAI: faster, better intent understanding, free for all tomorrow. Can be used in Hermes agent via existing subscription. ### Conclusion The AI landscape is rapidly evolving with Fable 5's potential return, new models like Fugu Ultra and GPT-5.5, and ongoing SEO updates. Building flexible systems like the Agent OS allows users to adapt quickly and automate effectively. ## Transcript We got some new updates today. So, it looks like Fable 5 might be returning pretty soon. So, apparently, according to Claude Code V2.1.190, it introduces several string changes that hinted preparations for a Fable 5 return. So, you can see here, for example, this is a tweet from Leo. Basically, if you look at the old one versus the new one, there's a bunch of changes inside the code that say Fable 5 and reference Fable 5 and everything like that. And that's on the new version of Claude Code. So, it's pretty interesting to see this. I think we're looking at Fable 5 coming back pretty soon. Um, here's another one. This from Chris GPT. Claude Fable 5 is showing again in AWS Bedrock documentation as well. So AWS they hosted Fable 5 before and it's now listed in two separate places which is the new version of Claude Code and AWS Bedrock. So I would expect it to come back pretty soon. You might be looking at you know the next potentially like 3 to 5 days. Usually when you see stuff like this, it is about to drop. So, who knows? But I would say Claude Fable 5 will be will be back online pretty soon. It looks like all the signs are there. It's just a matter of time at this point. Let's see who we got here. We got a new member. Thank you very much for joining. Randall, good to see you here, sir. All in the AM. Been looking forward to this one. Thank you very much. So, yeah, it's looking good. But I think um Fable 5 should come back soon. Let's have a look here. It looks like it's still down, but who knows when it's going to get released, but if they're adding it into the code in two separate places, I think it's just a matter of time, which is good because, you know, I think a lot of people worried it's not coming back at all. It's also a new version of GPT 5.5 instant. AI and the A. There we go. Sorry, I mis I mispronounced that completely. Oh, look at me. All right, we're going to kick it off with uh building out the agent OS. Thanks for joining in. Appreciate it. Let's run through the latest news from today. We'll consult the Oracle. So, we got the Hermes Oracle here. And you see, basically what it does is it pulls in the latest news automatically. So, this is running on schedule every 24 hours. If we take a look here, for example, we've got Hermes Oracle and then we have the news headlines over here. These are all based on trending tweets. So let's see what we got here. We got open eye unveils alapino AI chip. So, OpenAI is working on a new custom inference ASIC for chat GBT codeex and agents. What's actually really cool about this is we can draft a social media post with the Hermes Oracle inside our agent OS and then we can quickly create social media content about it as well. Let me show you what that looks like in a sec. And then we can also publish the article to our WordPress websites as well if you want. This system is inside the AR profitable boardroom. So if you look for example, this is the tweet originally. And then if we quote that, plug that in. Boom. Shaka. How easy is that to automate X? And if you're wondering, does that actually work? Check this out. It's crazy. Yesterday we hit 1.18 million impressions in one day from X using this system. So I know it's super powerful. And then we can also publish articles here. So if we want to quickly publish an article about that topic, we can just publish it like so. Pretty simple. See what else you got here. Bit of drama between Ali uh Alibaba and Anthropic MongoDB Groc plugin. There's a lot going on today. And then we can see everything that we published over here. So if we have a look at this. There we go. That is my Clawude hat indeed. When we've got exciting Clawude updates, I wear this. 5:14 p.m. in California. Welcome here. Baba, hey bro, love your content. My top guy to keep up with everything. Keep it up, G. Thank you very much, sir. Cyber, what is the exciting Claude update? So, we've seen Fable 5 pop up in two pieces of code today already. So, the new version of Clawude Code and also AWS Bedrock. So, it looks like Fable 5 will get re-released pretty soon. I would expect that to come out soon hopefully. It's all looking positive. So, can't wait to get that back. Do you know that's a really good point. It's like if it gets re-released, it might be US only, you know. I didn't even think about that, but yeah, you could be right there. I have a 4080 Super. What model would you recommend to run on my PC locally that I can get the most out of? Uh, probably Quen 3.5 3.6 I think for local models. Yeah. So, if you want to run local models or you can get free APIs, too, but I'd probably go with local models if you've got a good setup. Uh, yeah. And again, like Quen, Quen is probably your best option there. Hola. Heat. Heat. Hey, hey, hey. Awesome. Thanks, bro. Getting his cardio. And wait, that's the best way. Honestly, that is the best way. Usually I do I mean yesterday I did 21,800 steps whilst doing stuff like this. I think it's the best way, mate. Honestly, it's great for energy. Yeah, I would agree. I think Elon said Q1 next year. The GLM founder said it won't take that long. So either way, we'll get better models. It's just a just a matter of time. They're going crazy on open source models. It's mad. With or without wig? I mean, I'm wearing the wig all day. YOLO. Yeah, of course. Yeah. Yeah. Yeah. Try code. But what we did is we plugged in GLM 5.2 into our agent OS. So, if we go to GLM 5.2 over here, go to the workspace. You can see that we have uh GLM built in and we can just use it directly there as well. So, it's basically like using clawed code except it's a nicer UI and we can use it whenever we want. So, I'd recommend that instead. Superpowers, mate. Vibe coding superpowers. That's what we get. Yeah. I mean, 28,000 steps. Yeah. Yeah, for sure. Yeah, the wig helps you uh helps you with that for sure. GLM 5.2. I mean, you can just use it on CLI, so if you I I don't host it locally. Like for me, if you're using a Mac Studio, I don't think it's particularly good for local models. So, I think the best way really is just to to use it via CLI. You can get it on the coding plan and then you can build with it directly there. Heat. Heat. That is the the hat of champions. That is what I'm wearing right now. Is it comparable in price to claude? Why? It's much cheaper. Much cheaper. Exactly. No API. Plus, you can just subscribe. It's flat. I mean, is good for local models. If you if you want cloud models, I wouldn't I probably wouldn't use O Lama, you know. Uh but if you want local models, like it's free. So you you don't have to ask if it's worth it or not because it's free, you know. No, I'm not in England right now. Exactly. Exactly. What avatar is going to wear this hat? Welcome here. Me too. We got banter on the show today. This is good stuff. I like this. All right, let's see what we got here. Can you create a new guide based on the new updates that we did with Sakana Fugu Ultra and just compare it side by side with the other tests he did on Opus 4.8 and also Fusion. And then what I also want you to do is just create a full guide showing whether Fugu Ultra achieves fable five level intelligence or not based on our test. I know some of them didn't work as well. So just mention that inside the guide and obviously reference gold bench as well include plenty of screenshots, plenty of showcases. I think we created like 42 projects with Fugu Ultra. So we tested it a lot which means that you have a lot to show inside the guide as well. I've been doing everything you've been doing since G5 on two days. Wow, that's awesome. Thank you. I appreciate that. Yeah, we're up to Let's see how many members we're up to now inside the profit boardroom. It's growing like really nicely. We got a great community there. 3,800 members inside the community. It's amazing. So, yeah, it's been growing nicely. It's been a good journey. Graphy, I I looked at it and honestly, I couldn't see the value of it that much. Uh I think it's just for code mostly. It's not like Obsidian for example where you have a memory for everything. And it's more like you can just visualize code a bit better. It seemed kind of like a fad. I don't know many people who use it dayto-day. It's going to be interesting to see what comes out with 5.6 next, whether it'll be Fable 5 level or not. There you go. There you go. Yeah, I think it's coming. Just a matter of time. Yeah, I'm going to do a full tutorial on it, but basically if you rewind, you'll see at the start we talk about it. Looks like it's coming back. never missed a model so much. Same actually. Yeah, I'm looking forward to building some cool stuff with it again. Can't wait for that. Could you talk a bit about the memory tool that I can use for Claude 4.8? I built a similar agent OS. Just haven't added memory yet. currently using the claw vanilla. Yeah. So, I use obsidian. Obsidian is pretty cool. And uh what I call it is the infinite context engine. Right. So, when it comes to memory, we have a system where you've got for example OMI and OMI is taking notes on me each day, right? So, it sees what we're working on, sees what we're doing etc. Um, and what it can do is it can export all the notes that it's looking at from your microphone and from my screen as well. And we can sync those to Obsidian every day. So, it adds a markdown file with our memories. So, that's part one of it. Then part two once it's done that is inside Obsidian you've got your full memory list as you can see right here. And so, if we go to Obsidian, it gets updated daily. And this is like a full database as a second brain of everything that's going on in my life. Right? Each one of these dots is is a separate markdown file with different memories and notes about me. So that's part two. And then the most important part is that you plug it into your AI agents. So we have a memory galaxy that we've visualized as you can see right here. And this is plugged into our agent OS where we have for example we have like Claude, we have Hermes and they can all run and look locally at what we're using inside the memory. And then every time we use AI like you see this is just updated right now. Every time we use AI this is logged to our system. So it's automatically updated and then our AI agents can organize that into a beautiful system like this. And so number one your agents can use this memory for context. But number two, they update the context for you. So you don't need to I've never touched this. I've never written anything inside the memory galaxy. It's just automatically updated by my AI agents. So that's the system that I recommend for memory. Before so I was I mean before like getting into AI and I still do I have a SEO agency. So before that I had an SEO agency and that's how I naturally got into AI because obviously you could use AI for SEO content and even to this day I still do a lot of experiments with AI and it's it's really good for SEO. So we still have an agency that's big part of what I do. And then you know we're still building out websites as you can see right here. This website with AI SEO, it went from like three clicks a day all the way up to 278. And this one zero all the way up to 72 clicks a day. So just, you know, I was running a lot of SEO experiments, running an SEO agency, got a team with that, a lot of clients, and that's what we still do today. Yeah, 4.8 is still good. Don't get me wrong. Obsidian is amazing. 100% agree. Set and forget. Exactly. Is Obsidian open source? Yeah, it is. Also, it's free, so you don't, you know, you can mold it and change it however you want. Uh, for that, like I I don't know. I'm not in those niches for our clients. We just use normal SEO systems and and it works really well. We don't really worry about stuff like that. All right, we're getting this set up. Yeah, exactly. Yeah. Yeah. We we we use AI for that, right? So AI for the content, AI for the blogs. It's pretty powerful stuff. I mean, I'll give you an example. Even this morning when we were consulting the Hermes Oracle, let me show you Hermes Oracle here and based on the latest news, we can just publish like an article about one particular topic. So if we have a look here, for example, we have this article that's published on this new trending topic. Um, I didn't need to log in. The AI writes it for me. It looks beautifully formatted and we can easily create content on trending news. Plus, we can get it indexed with index optional as soon as we create it. So, we still create content today. Uh, and we do link building like our main thing is really link building, but I run a lot of experience with content as well. Yes, I do. Um, well, I have a team that does that for me, but yeah, we have a lot of training on that inside the aircraft for boarding. Yeah. So, the the setup is I mean it's very simple, right? is you post you you get OMI to take notes on you those notes go into Obsidian then Obsidian goes into agent OS and that forms number one the markdown files and number two the knowledge graph automatically so it's it's really simple as a system very easy to like you know set and forget it basically Heat. Heat. Oh, me. Yes. So if you go to the air puff for boardroom here and then what you want to do from here is you go to the classroom and inside the classroom go to updates and it's the infinite context engine. That's the system that you want. Shows you how to use your agents with OMI and Obsidian. Pretty powerful system. Oh yeah. Yeah. So there's the authors of course. Yeah. Yeah. So whoever owns the site, we add it inside the website, right? So for example, if we go to the air profitboarding.com, you can see that I'm the author, right? So we add information about me. Plenty of eat on there as you can see. Let's see what we got here. Awesome. Let me know how you get on. It depends what we're doing it for, right? But we have a system for generating them with this playbook here, right? So, if you go to this section, we show you exactly how to create that type of content. And here too, we have a full section on Instagram as well. So, if you want more stuff on that, if if you like that sort of stuff, we've got loads of tutorials on that right here to help you based on what's working for me. Um, now what what we use now for SEO, I'll show you. So, you got the agent OS and then if we want to do keyword research, for example, we can just research what we're working for, what we're ranking for, and then plug in the keyword here, right? So for example, we can say okay, hey, profit boardroom last 28 days. Let's type in Hermes. And what it would do is research for our Google search console data. And that's how we do the keyword research. We don't really use HFS anymore. We look at okay, what's our data? And then how can we target keywords around that based on what people are searching and finding our website via anyway. So for example, like Hermes change log, that's a pretty good one that we could create content around. Hermes release notes. These are all new keywords that we haven't gone for, but we could go for Hermes agent plugins. That's pretty cool. Um, and it ranks the content by the opportunity and you got, for example, the amount impressions, the amount of potential clicks, the CTR. So, you know, okay, if the CTR zero, then what we could probably do is create a new page around that targeting that new keyword. So, that's the way that we look at it. And then we can do that for any of our sites. It's pretty simple and easy. I actually want to add more websites in here. I'm going to do that now. And then from there, we can just click on use that topic and deploy the content and automate it with the agent OS system. As you can see, back in a sec. Heat. Heat. Yeah, I'm back. Back in the game. No. Why would you do that? So, it's leaked that it's coming back. They said in FA for Fable 5, basically there's two leaks. Number one is that it was found in the code on AWS Bedrock, which is a sign it's coming back. But then the other one is that it was found in the new version of Claude Code as well. So, there's two references, two references to it in the latest two updates from AWS and Claude Code. Thank you. Good day to you too, Serge Jay. Good to see you back. Started using obsidian when it first started with the nose and stuff. Glad to see they're evolving. Yeah, it's really good. Julian is using Fugu as a rebuild in his tool. Something like that. See what we're up to. I would I mean it looks like it's coming back before Uh, I would say early July or in the next 7 days, but let's see. Yeah, early July and the next 7 days I would expect it back. But you just never know. I mean, it might not even be available outside the US as well. So, that's something to bear in mind, too. No, I'm based in Asia. Yes. Originally from Manchester. Just having a look at this tweet from Hassanto. They open sourced a long memory system for AI that reduces tokens by 61%. Let's have a look at it. It's from Tencent. What a genetic coding environment do you suggest? GS. So for us personally, we use the agent OS for coding and we use CL desktop as well to build out the agent OS. So I would recommend that cuz you can have like multiple different agents working together. You can orchestrate them and if you want to actually like code and build stuff. Let me show you an example. So like we were using uh Fusion lately and we can just see everything that we've created and edit it whenever we want to as well. I think that's a really cool way to use it. So you know if you want something where you can orchestrate all your agents together then I would go with something like the agent OS if you want to build something like the agent OS use claude and then if you want my setup it's inside the air buffer boarding for the agent OS system that I use personally. Yeah, I haven't really that that's not something we Okay. Wow. This is pretty wild. Fusion came I mean Fugu came up with a lot of good stuff. Yeah, exactly. This is with Sakana Fugu Ultra. You can go to sakana.ai. But if you want the Yeah, if you want to see like the full guide on each of them, go to Goldiebench. So if you go to Goldie Bench here, you'll see all the details on like every single every single model, etc. what we've built with it. You can test them all out yourself as well. Actually, we should add the website on each as well. On each model, can you add the official website as well and an overview of the full details, but just make sure like you actually fact check it, update that now for each model that we've tested. Yeah, it's available at sakana.ai. Uh we've already built it into the agent OS. So, if you go to Sukana Fugu here, you can see everything that we built. Um, and it is good. It is good. People I've seen like mixed results online of like people testing it, but we tested it relentlessly across 42 builds and it looks good. Like it can create some awesome stuff. Yeah, it's pretty uh just just straight out of the box. We give it we give it the same prompts to everything and then we build it straight out of the box with the prompts. Stuff like Sakafugu is one shot anyway. So you can't go back and forth with it or teach it a skill or anything like that really. You could, but it would be a it' be a lot of you use a lot of tokens on the replies in the context. All right. Uh, this interesting Okay, nice. Create a full guide on this and then add it to the workspace inside the agent OS. create a bunch more as well in a separate folder in the workspace so we can see them side by side. Probably want about 5 to 10 What is it using to build with? What do you mean? Like what's the I mean, it's just using HTML, right? I don't know. I'm not I'm not that technical, so maybe there's something I'm missing there. Look at the difference between this and this. Crazy. The way you describe that. No. I notic in the latest updates it didn't find the new Fable 5 leaks in the Oracle. Why is that? Do we need to tweak something there? Because that's trending right now and I thought that would be like probably the most important thing. Let's go. Thank you very much, sir. Yeah, I think you'll like the Agent OS. Totally understand. But I appreciate you joining is Fable 5 coming back today. We're looking at this. So, we have uh a couple of interesting updates on Fable 5 that's going to basically determine, okay, is Fable 5 coming back or not? And one of the biggest updates, you see this tweet from Leo right here, is that Claude Code V2.1.190 introduces several string changes that hint at preparations for a Fable 5 return with it being included in subscriptions as well. So, I'm going to walk you through, okay, is Fable 5 coming back or not? What can we expect? When can we expect it? What does it mean? And what are the new leaks on this update? So, let's kick it off here. First of all, nothing is confirmed. These are just leaks, but the signs are there, the signals are there, and it's looking pretty promising. And obviously, Anthropics Fable 5 got taken down. Honestly, from what I've seen when I've been testing models ever since, like Opus 4.8 number one is nowhere near and a lot of models are actually beating Opus 4.8 by a long way right now. We've tested out if you you can test out and see all of our tests on on Goldieb. You can actually see what we've created and how it performs if you want to. But the main point here is like nothing I is really as good as Fable 5 yet at this point. And so everyone is waiting for Fable 5 to come back. So what's happened? Well, on June the 9th, Anthropic shipped Claude Fable 5. Um, and then on 3 days later, Fable 5 got taken down, right? And then on June the 24th and 25th, the leaks started showing up. You might say this is just a rumor. You honestly like every time I see stuff like this, it's usually a pretty clear signal that something like this is about to drop. And we saw this with for example like even Opus 4.8 itself before before Fable 5 came out before like usually when you see this sort of stuff showing up is it's about to come out and and I'll talk about time frames in a minute as well. So, there's three separate checkable trails. This is not just like one rumor that's dropped. And the three checkable trails, which I'll run through each in a second, a clawed code and an update, the Amazon Bedrock catalog, and the AWS docs. And they're all moving in the same sort of time frame, which suggests, okay, if three things change across three different releases, something is happening here. So, let's walk through each one. What's happened? You can see the timelines here. So June the 9th it got dropped then it got pulled then all the signals coming out on June the 24th to 25th. So the first one is Leo was digging into claude code v2.1.19 and found the string changes. The old line fable 5 purchase separately from your plan was actually removed and the new line was you've used your Fable 5 usage for this week added. Right? So this is add added inside the strings. So that kind of hints at something big that not not just Fable 5 is possibly returning because why would Claude add that otherwise right? Why would they change the string? But it may also be included in your subscription with a weekly allowance in in terms of your tokens that can be added here. And you can see like the before and after on this tweet like so. All right. This is the old version. This is new version and there's a bunch of new updates regarding Fable 5 on the new version of Claude. Next up, one of the biggest news accounts, Chubby, he picked up as well. Usually when he's talking about new models, they get released pretty quickly as well. And so, not only that Fable 5 is hinted at returning inside of these things, also it suggests the model may be permanently added to the subscription plan, which would be great. I mean, for me personally, you know, when they said, basically, when Fable 5 was announced and released, what they were saying was it was going to be inside the subscription until June the 22nd and then it would be taken out, but it sounds like they're actually going to be adding it back in as well. Now, also, there's a stronger confirmation, which is it's in the catalog. So, obviously like strings are a bit of a hint, which is what we saw with Claude Code, but this is an even bigger um update as well. So, Bridgemind actually spotted Fable 5 back in the Amazon Bedrock model catalog as well, which is pretty interesting. So, the page is live. The model is listed by Anthropic. And now you got two separate trails, which is the Claude code strings and a live catalog page both pointing in the same direction, which is saying, you know, Fable 5 is coming back soon. So, you can see the tweet from Bridgemind here. Um, he says the page is live. the models listed um by anthropic combined with the Claude code updates. The signals are there, right? The Fable 5 is coming back pretty soon. And you can see he's posted the example page right here from Claude Fable 5 on Amazon Bedrock. So, just to be clear here as well, there's three updates on this. Number one is the claude code strings. Number two is that Fable 5 will probably be included inside the subscription when it's back. Number three is that it's been it's reappeared on Amazon Bedrock as well. Also, there were a lot of hints at Sonic 5 coming out recently as well. So, that's pretty interesting, too. And also Chris Chris GPT he was saying that Claude Fable 5 is shown again an AWS bedrock docs and model cards as well. So AWS lists the model life cycle as active with bed rock ids including anthropic claude fable 5 and global anthropic claude fable 5 as well. So let's have a look at the tweet. He screenshotted it here as you can see here which is pretty interesting too. So all the signs are there you know everything is looking like Fable 5 may be back. One part to note here is you know is this only going to be in the US? That's that's another thing to note. So for example if you're outside of the US obviously originally when Fable 5 was taken down it was taken down for people only in the US but the problem was Anthropic couldn't really moderate that and so they just took it down for everyone. And so when it comes back, does that mean it's only going to be available in the US or will it be available for everyone? Who knows? And then when can we expect it to come back? It's going to be interesting. I think we can probably I I would expect this to be back by, you know, early July or within the next 7 days. But it's all looking really positive. You know, either way, whatever model comes out, whatever happens next, you really want a system where it doesn't matter what model comes out, and you focus on the system, not the model. So, we have the agent OS, for example, and like the day that Fable 5 got taken down, we removed it from our agent operating system. The day Fable 5 actually came out, we added it inside, right? And so my point here is like it doesn't matter what model comes out. Just have great systems that are super flexible to whatever comes out next. I'll show you another example. So when we were using uh FA Sakana Fugu that just came out a couple of days ago, as soon as it dropped, we'd already added it and created stuff with it inside the agent operating system. As you can see right here, we even created a an a separate operating system called webOS with Fugi as well. And so my point here is like whatever drops, you want to have a great system to organize it with. And that's exactly what we've built with the agent OS. As you can see right here, if you want to get this, it's inside the AI profit boardroom. If Fable 5 gets released, we'll have new guides, new tutorials. We'll add it back inside the agent OS as well. And inside the community, you can ask questions, get help and support whenever you want. Inside the classroom, you can see all of our new trainings and new daily updates. Inside the calendar, you can actually jump on four weekly coaching calls about AI automation, how to save time, how to scale scale. Uh if you get stuck with anything technique, you can ask questions inside there too. Inside the map as well, you can meet people in your local area who are building with Claude and other models like this. And if you want to get all that, it's inside the air profit boardroom link in the comments description or go to the aiprofit.com. Who we got here? Banu, welcome here. Sharon, welcome here as well. Good to see you both. So, Sakana Fugu Ultra has been out for a few days and this is according to benchmarks able to achieve Fable 5 level intelligence. Now, does it actually match that standard? How does it perform? Where's it up to? Is it better than Fusion as well? I'm going to cover all of that in this video today. I'm going to show you what we've built with it. We actually built out 42 different things with Fugal Ultra, so you get a really comprehensive idea of how powerful it is, whether it's worth using, the best ways to use it, and we've tested it on Goldie Bench as well. I'll show you the benchmarks and the leaderboards in a second in terms of how it performs versus everything else. also how it performs side by side and what we've built with it so far. So we have tested it absolutely relentlessly using our system here. So we have Sakana Fugu plugged into our agent operating system like you can see and then what we can do with this is we can build cool new stuff with it and then test it out side by side. The only problem was that when it first got released which was just a couple of days ago you were kind of limited by the number of tokens you could use. And so the problem with that is like number one, it was super slow and number two, you couldn't really test it as much as you can without getting like locked out for 5 hours. So we've built 42 different things with it. Today I'm going to just guide you through what it is, how to use it, and also how it performs versus Fable 5 Fusion and Opus 4.8. So let's get straight into this. We ran 42 builds to find out. And we use the same 42 prompts with three models. Fugu Ultra which is from sakana.ai. We've got Fusion which is from open router. Both of these on benchmarks are claiming to achieve fable 5 level intelligence. And then we have Opus 4.8 which is the only or sorry the most powerful model you can get from Claude right now. And we'll compare these side by side with our Goldie benchmarks. And we've tested it, you know, with with real builds that you'll see in a second right here. By the way, if you're wondering, okay, what is Sakana Fugu? How does it work? So, it's it's basically a way of orchestrating multi- agents together, right? So, you have Sakana Fugu, you have closed and open models, and that's an LLM pool that basic basically fuses an answer together. So, the idea here is like if you have multiple agents working together, multiple models, closed and open together, then you tend to get better results than just one single output from one single model. And so on their benchmarks, they claim to achieve better level intelligence than Fable 5. So you can see the benchmarks right here. And this is compared against Mythos preview for example. And you can see Fable 5 here too on the benchmarks. And on a lot of the benchmarks from this Japanese AI Fugu, which a tiny little AI lab, uh they are outperforming Fable 5 right now. Let's see how it performs in reality on the actual builds that we've created right here. So on the build, I'm going to be 100% honest with you on our leaderboard with Goldie Bench. You can see here that Fusion is actually top right now. It's actually crushing everyone else. Fugal Ultra is not doing too badly, but just have a look at the builds yourself today and see what you think. Um, and see what you think about in terms of what you can get out of it. Now, one thing to note as well is like if you're using Fusion, which is from Open Router, follows the same idea. Use multiple AIS together. With Fusion, the difference is that you pay per API. With Fugu, you use a subscription. And so with the subscription, it's a flat plan. With Fusion, you you pay per token. So that's one of the biggest differences. Also, Fugu Ultra is from Japan, whereas Fusion is from Open Router, which I think it's US. And also with Fugu, if you're on a flat plan, you can obviously run out of tokens just like you would with a Claude subscription. Whereas the difference with Fusion is that with Fusion, you don't run out of tokens because you're paying per API, if that makes sense. So, let's test them out and see what we've got here in terms of builds um and how they perform. So, first of all, we've got the solar system test right here. And you can see that we have Fugu Ultra, Fusion, and Opus 4.8. So, we tested all three together. So, this is the version from Fugu. This is Fusion and this is Opus 4.8. Now, if I had to pick one, I'd probably go with this one. I think it's the most sophisticated and elegant version. I do quite like the output from Opus 4.8. I would say that's coming in second. And then Fusion comes in last on that particular example. Let's have a look at the next one. So, this is a Voxil Temple run, like a runner game. As you can see, pretty nice outputs from Fugu again. and it can build some super nice stuff. As you can see right here, uh, if we compare that versus Fusion, I would say probably Fusion's created a better example. And then if we compare this versus Opus 4.8, wow, that is bad. It's just so basic, isn't it? It's just not very useful at all. Now, I will say the most frustrating thing about testing Fugu was that, you know, we just kept getting locked out for 5 hours. So, so bear that in mind. It's like it's pretty cool. It's creates some awesome stuff. However, not that great if you want to test it a lot, if you want to use it a lot. Um, yeah, it it just, you know, it would take sometimes 20 minutes as well for a response and then sometimes it would fail as well. So, that's something to bear in mind. Also, one thing to bear in mind with like both these setups, for example, Fusion and Fug Ultra, is that they're one shot because you wait 15 to 20 minutes for a response and that's it. you know, you get multiple answers, but you're not going to go back and forth with it like Chachi Bet or something like that. So, this is the output from Fugu Ultra. This is Fusion, which is a bit laggy, but not bad. I would say Fugu's game was better. And this is Opus 4.8, which kind of just feels a little bit retro for me. It's like not even full screen. It's probably it's slightly smoother than Fus uh than Fusion's output, but I would say again Fugu Ultra is winning. So, you know, on all the tests so far, pretty much Fugu is is winning on all these. Let's have a look at the next one. So, this is a fireworks display. Um, basically, and what you can do is like you can click and to generate the fireworks and then it will follow you around based on where you you put your mouse, right? Which is pretty nice. So, it looks really cool. Let's compare that versus Fusion. Fusion just literally didn't work. Um, and then we have Opus 4.8. So, Opus 4.8 is outperforming Fusion, but it's nowhere near the same level as Fugu. Next one. This is uh a Nordic dungeon crawler. So, let's have a look here. Wow, this is pretty nice. This is from Fusion again. I mean, it looks super nice, right? Nice 3D sort of setup walking around this Norwegian dungeon. Uh, super nice. like 3D style. I like that. I like that. It feels huge as well. I don't know where this world ends. That is one of the things I found with like when we're creating games with, for example, like uh Fugu, the the games feel like endless, if that makes sense. They feel absolutely huge. So, let's just go over here and see if we can get those coins. See if actually Oh no. Oh no. Sunshine. Who's this guy? All right. So, yeah, it's pretty cool. Now, let's have a look at that one. So, this is from Fusion. So, I mean, I don't really know what's going on here, to be honest with you. Like, it's very difficult to see anything or to understand what's going on, right? It's just like it's so dark. I mean, it looks it's kind of cool, but I don't know. I literally don't know what's going on there. And then we have Open 4.8 where you can't actually get through the start screen, right? It totally failed. So, you see the level difference here. you're seeing how different it is between each of these models. This is not like, for example, like a a a close race at all when it comes to to being compared to Obus 4.8. It is clearly at a massive massive um level above Opus 4.8. And that's why we would say it is comparable to Fable 5. you know the I wouldn't say it's as good and I wouldn't say you can go back and forth with it as much but it can create like comparable stuff as you can see now this is an open world game from Fugu as you can see this one just feels a little bit more basic to me like I think you need better graphics again the problem with this is like you can't go back and forth with these games so if you create something that's it unless you get Claude Opus 4.8 8 to tweak the codebase later, you're kind of stuck with the outputs. Whereas, for example, if you create something like this with Opus 4.8, you could say, "Okay, change this, change that." Right? It's not just one shot. Let's have a look at Fusion. Fusion's Sorry. Yeah, Fusion. This is even worse, I would say. Super basic. Feels quite slow when you're walking around. The graphics are not nice at all. And then we have Opus 4.8. And again, we can't get through the enter screen. Like, it's lally just stuck there. Opus 4.8 seems to do that a lot. Let's see as well how it performs on the benchmarks versus Fable 5. So Live Code Bench outperforms Fable 5. SWB Pro, Fable 5 is outperforming Sakana. And then on Terminal Bench, Fable 5 is outperforming uh Sakana as well. So, just to be clear here, like Fugle Ultra wins on Live Code Bench 93.2 versus Fable 5's 89.8. It tops GPQA diamond at 95.5 above Sakana's own Mythos preview number right at the saturated frontier and then it trails on SW approach. So, it is pretty close and again it's it's a I mean, Sako is just crushing crushing over 4.8. Even Fusion is as well, but from what I've seen, Fusion is is not as good as Fugu Ultra when I test it. Also, something to bear in mind here as well is that there's Fugu Ultra and there's Fugu Mini, right? So, if you want to see some stuff that we've created with Fugu Mini, let me pull this up and we'll have a look at the models here. So, this is Fugu Mini. So, it created some interesting stuff. Now, if you want to see it versus other models, we can go to the compare section here, and we've got head-to-head comparisons of all these models. So, if you want to see Fuga Ultra versus Fugger Mini, we've got that right here. We can see how it performs. So, you know, Fugu Mini is a cheaper, faster version of Fugu Ultra. So, this is kind of like a flight simulator game as you can see from Fugu Ultra, which is quite hard to control, but pretty fun. And then we have Fug Mini here, which actually feels a bit smoother to be honest with you. It's not that not I mean the graphics are not as nice, but it does look like more pleasant on the UI. Got a couple of examples here. So, we recreated the same game with both. Now, it says you can use WD to move, but you actually can't on that. And then Fuger Mini literally can't get past the uh start screen here. That was the other problem as well we found was like sometimes the tokens run out in the response. This is Fugu Ultra which is pretty insane. Look at that. Wow, that's pretty nice. If we look at Fugu Mini, not quite as good. Look at that snow. That snow is nowhere near the same standard as as Fugu Ultra. So there's a huge difference even between those two different models. I think Fuga Mini is kind of like Opus 4.8 for for Sakana basically at this point. So yeah, there's a big difference between them as well. And overall, I would say, you know, if you want something that's just a better step up from Opus 4.8, yeah, for sure, I would go with Fugu Ultra. Like, it it performs really good. It's created some awesome stuff. Um, what we've actually done is if you want the full setup, you can get that inside the AI profit ward. So this full agent operating system that you see here with Sakana and Fusion built in. So you can chat to either of them and you can see what we've built inside the the workspace here. So everything you create gets saved inside the workspace too. This is all inside the AI profit boardroom. And this is an agent operating system where for example you've got a memory plugged into every agent you use. You can orchestrate all your agents with paperclip and a group chat. You also got a pipeline where you can go from idea to implementation really quickly. And we have a camb board for orchestrating your agents too. We've even for example built out some awesome custom stuff with Hermes agent include the um Hermes Oracle and all sorts of cool stuff for outreach and lead generation tool as well for this sort of thing. So if you want to get that that's all inside the AI profitable room link in the comments description or just go to the apiprofit.com you can get the agent operating system with fusion sana and claude all built in along with Hermes right here. We update it daily and you can see the last update date. We have a video tutorial and a guide on how to install it and with the resources over here as well. And then we drop new daily tutorials based on what's just been released. Inside the community, you can ask questions and I create video tutorials helping you every day on everything that you ask. And then inside the classroom, you can go from beginner to expert in just six weeks with our course right here with AI. And inside the calendar, you can jump a weekly AI automation coaching calls. Ask questions, meet the community, share your screen, ask questions about your setup as well. Inside the map, you can meet people in your local area who are building with AI agents. And that's all inside the AI profitable room. Link in the comments description or just go to the AI profit.com to get access. Thanks for watching. Yeah, that's it. That's why we test both. Sheena is here. Welcome back. Wow, nice. Sounds like you're making awesome progress there. Congrats. Is that for your first client, Sheena? All right, that's awesome. in the same day. That's a lot of traveling. Today we have a brand new update from Google as they have released the June 2026 spam update and this is just when things were settling down because they've been releasing all sorts of crazy updates this year and this is the first one for the spam update as you can see. So I'm going to talk you through what it means what it is what it means for your websites if you're doing SEO and how this is set up. So, this just dropped on June the 24th, and you can see on the Google search status dashboard. Usually what they're trying to do is just wipe out some of the dodgy stuff people are doing with SEO, particularly with AI and that sort of thing. So, what they try and do is like basically they have these automated systems to detect stuff, you know, and they're constantly like just running those in the background. However, when they do something like this, they basically improve the whole system and so this can impact your rankings dramatically. Uh sometimes, for example, some websites get totally wiped out. Uh sometimes they do link spam updates as well to remove like dodgy links that people have been getting off, you know, directories or whatever it is. And sometimes as well, it can actually make an improvement to your site because your competitors go down too. So, if you want to see the guidelines for this and when it is rolling out and when it actually stops, you can see the Google search status dashboard. And every time there's something live that's actually impacting rankings for SEO, you can see them right here. There were actually rumors that this has dropped before and actually dropped out a few days ago. Uh, usually as well, I think these take about a couple of weeks to roll out and see the full effects. So if you see your traffic going up and down, that's pretty normal. You can also see there's been a lot of updates recently. So for example, there was the Google discover update on the 5th of February. Then there was a March spam update 24th of March. Actually, that's this is the second one this year. Sorry. Uh yeah. So there was one in March, 24th of March. And then on March 2026, we had the core update. And then on 21st of May, core update. And you can see the last spam update from Google lasted 19 hours 30 minutes. It was a very quick one. Whereas the one last year in August took 26 days. So who knows how long this one is going to take. Sometimes it's weeks, sometimes it's days, sometimes even hours as you saw before. Um, also this applies to all languages and rolls out globally and it may take a few days to complete as mentioned here. Now you might be wondering okay what happens during these sort of updates. So basically Google systems always working like 24/7 right but once in a while they ship a big upgrade to those systems and that is what this is about. So the June 2026 one is the latest. The engine behind it is called spam brain which is Google's AI based spam prevention system. Each update makes it better at spotting old tricks and catching new one. So number one, it's automatic. This is not like a manual thing. No, if if you get, for example, penalty or your rankings drop down, it's just like an automatic algorithmic thing. And number two, it looks at Google's spam policies and then it compares everyone's sites against that. Bear in mind, it doesn't always get it right. A lot of people have been wiped out this year for like no reason. I'll show you an example here that was pretty interesting. So, if you were ever familiar with uh Brandine and Back Lingo, his traffic actually got crushed. And uh actually the site was sold to Sam Rush I think it was. So it's not him that runs it anymore. But the point here is like you know even when you got really good content sometimes you just get crushed and uh that's part of the game. One thing that we do to try and diversify away from that is we have an agent operating system where we can post content to multiple websites. And so you know if one does go down and that can happen and sometimes it's not your fault, no problem. you've got multiple other websites ready to go where you can publish content and you can start ranking and you're not starting from scratch again. So that's the way that I look at it and I think it's a good way to kind of diversify. The same with social media like we'll post content across multiple different social media platforms because you just don't know like what can happen to your main websites. So these are the policies right now and they talk through like each of the things they do not recommend because this is what they measure your site against. One thing to not as well is like Google's not always right, you know, like they sometimes they make mistakes like we all do. We all saw that with for example like the helpful content update a couple of years ago. Loads of of good websites just wiped out and um and that was brutal. That was a brutal time. So just something to bear in mind there as well. So some things that they go after like scaled content which is just like creating loads of pages, site reputation, uh like parasite SEO, expired domain stuff, dodgy links, and super thin content as well. Now, you also might think, okay, does that mean I can't use AI to write content? Using AI isn't the problem. Google has said this plainly, like how content is made doesn't matter. Where it's helpful matters. An AI that helps you publish like genuinely useful original pages is fine. So, I'll give you an example of this. Like, if we have a look at this website right here, this is Goldie Bench. And this is all like first party data. It's all information and experiments that I've personally created with SEO. So this is all like kind of documenting the work that I do. But I've never logged into this website. It's all created with AI. And basically cord just documents with our agent operating system what we've created and what we've tested. So my point here is like if you are creating content with AI, but it's adding information gain is actually useful. It's actually unique. Well, that helps a lot. I think there's nothing wrong with that at all. Now, if your traffic drops, check Google Search Console number one first, right? You can look at performance and then just check your impressions around the 24th of June. Obviously, that data won't be out yet. You just have to wait for it to roll out inside Google Search Console. Also, I would confirm it's the update. to check that the drop actually started around the update. You want to read the spam policies honestly. Just go through it and just ask, okay, does any of your part uh any part of your site cross that line? And then I would wait for the the update to roll out properly and then you can prune or rewrite or improve your existing pages, but it's going to take a bit of time, right? So like don't do anything wait for a couple of weeks after the update to make any changes. Don't do anything drastic and watch it roll out first before you make any sort of ideas. I think as well one thing that's really good is like it can be a bit emotional when your website takes a big hit in traffic. So you just want to create some space between that update and when you actually start making any changes and then you know sometimes it can take like months not days to recover from this sort of stuff. So you just have to to be patient and that sort of thing. Now, a few things to worth knowing. If if nothing drops or nothing changes on your website, you don't need to change anything. Using AI, totally fine. As long as you're creating good content that's unique. It does take months to recover from something like this. And sometimes it's not even your fault. So, just bearing that in mind. And now, if you want to get my best AI automation systems and I'll show you a couple of examples of what we're doing with SEO for our website. So, let me pull up a couple of examples right here. So that Goldie Bench website that I showed you a second ago, we just created it and is already getting pretty good traffic. Bear in mind like some people will be searching for it. So that's probably why. But it's doing pretty good in terms of SEO. And we just created this last week, which is crazy because it's it's gone from zero to 68 clicks a day. Again, probably people just searching for it, but just something to bear in mind there. This is another website we created that's actually performing very well with the systems that we've used. And here's another one as well. Now, if you want to get a free SEO strategy session where we look at your website and talk about what's working for us and then apply that to your website, you get a customtailored SEO game plan plus a link building strategy. You can get that at goldie. Link in the comments description or go to goldie. And you can book in a free one-to-one SEO strategy session where we can look at your site, give you advice, etc. If you want to get my systems for using AI SEO, we've got that inside the AI profit boardroom right here. So, if you go to the agent OS section, you can get our video tutorial. Um, the agent OS, you can see it's updated daily, the zip file to install it, and then we have new daily updates. And if you like AI SEO, we also have a full AI SEO automation section here as well. So, if you like AI SEO and that sort of thing, you can get the full course, community, and systems for that right here. Based on what's working for me, I think I'm one of the few people that actually tests and shows the results of what we're doing with SEO. So, yeah, interesting times. Let's see what happens next. We also got a free course link in the comments description. See you on the next one. You are real. I absolutely am, mate. Uh, okay. Yeah, that's it. That's probably a smarter way to do things. Zoom call. Not traveling all the way there. How many articles do you post daily to a website? Usually just one to two these days. Just like to focus on quality, not quantity. And Forbes has been using AI content for 10 years. Absolutely. Absolutely. Nothing wrong with AI content. It's just how you use and how you create the content that matters. me. I I do in a way and I don't at the same time. It's an interesting one. Mixed feelings about that. Nice. Good to see you leveling up there. Thank you very much. Have a great day to you too, sir. Thanks for all the questions today. Yeah, it's amazing. Uh what I mean, all right, so you are the creator says Hermes uses a lot of tokens. You can just use free APIs or you can if you for example like subscribe to Twitter then you can use a with Twitter and then either way you don't even need to worry about tokens like tokens is not even something I think about to be honest with you so it doesn't really matter on that point. Today we're going to be running through the latest questions on the agent operating system, how to use it, how to get the most out of it, etc. So, for example, you can see inside my system, we have Hermes Jarvis ready to go. It's a voice activated version of Hermes. We can build stuff out of it. We can see our history of this. Also got Hermes Oracle which basically takes the latest news for example like the Fable 5 announcements on Bedrock and then we can publish blogs to it directly to WordPress. We can draft social media content on it and we can quickly find the latest news every day and it's automatically refreshed, right? And so these sort of updates that just dropping all the time and these new systems that we set up, we can just improve this whole agent operating system in one shot, right? Um, and every day the system compounds, it improves, it gets better, and it's probably the best, the most fun way I've seen to orchestrate agents. We also have a team orchestration section here where you can link all your agents together. And today, what we're going to be doing is running through the latest questions inside the AI profit boardroom that we've got from our community about using Asian operating systems so that I can basically help you answer them. And also, if these people have these questions, if the community has these questions, then you probably have these questions as well, right? So, we're just going to run through the latest ones to see what people have. So, let's have a look. We got a post from Joseph today. He said, "I was skeptical of joining, but Julian's video is an excos from months ago. Only just joined a few days ago." and he said, "My advice to anyone entering this AI world, start with a base and then step by step build on top of that, which I 100% agree with." Right? And so what he's actually done is built out our agent operating system as you can see right here, but made it his own. So he's tweaked it, he's improved it, he's designed it how he wants it. He's got the council over here, he's got open claw, he's got the loop engine, etc. using the systems that we've created. And so my full agent operating system is inside the profit boardroom. And basically what it's done is tweaked that exactly how he wants and built out his own memory galaxy for Obsidian 2, which looks really, really cool. And so I 100% agree with this. Like you don't need to build everything out at once with an agent operate system. Just do it one step at a time. Enjoy the journey. Realize you're super early and then you can improve it from there. And I think it's just one step forward at a time. I built the whole thing from scratch. And when I built it, I didn't know it would evolve into this amazing system. But it's just something that you get your your head around. You you grow your skills with building it. And I do think it's a skill itself. You know, orchestrating and building agent operating systems is a skill and probably one of the most powerful skills out there. Joe as well has built out something similar. So you can see his setup and he's organized it slightly different. So he actually used our layout as you can see right here. He said, "I hope you don't mind, but I use your layout." Which is totally fine. That's that's what it's there for, right? That's why we share this stuff. So you can see my layout over here. The great thing is you can customize it how you want. So for example, the way that he's used it and set up is that he has systems, agents, projects, analytics, settings, and then he's changed it exactly how he wants right here. Um, and he's got his memory systems over here as well. So there's many different ways you can organize the actual UI and the setup, and there's no wrong or right answer. This is just a really good way of setting up your systems. I actually think it's really organized. I also like the minimal style that you've applied on the UI so it's easy to navigate to each section and you split it by agents and systems and everything else. So great job and thanks for sharing You can see for example like Sheena as well who's another member. She fully customized hers and you can see for example she's like added her own logo in here totally changed the name. She's got a command search which is pretty useful because then you can search across all the features. Like for example, we have like a outreach and lead generation system inside ours and a email campaign system. And what would be super useful is actually adding a search system. We have a filter here and we can we can like turn things off and switch them back on if that makes sense. So for example, if I don't want to see the local engine anymore or the ancient cambere, I can click on um complete and I can hide things off the menu as well to make it more minimal. But what would be really good is adding a search feature as well, like you can see right here. And then she set up a video agent system just like we have here for AI avatar videos too, which is pretty cool. So you can like basically in one click automate and create amazing videos as you can see with your AI avatar and it's you just like type in a prompt and the system will research it, create the avatar, edit the video, add the B-roll, do all the research um and plug it all together like you can see. And it's it's absolutely amazing what you can build with this stuff. The other thing I would say here is like whatever you're working on dayto-day. You know, let's say for example, you're working a lot on SEO. Just build out one workflow inside your agent operating system for that. And then go one step at a time. Because for example, if you build out that automation this week and then you set up another automation, for example, for videos next week, then this stuff just improves and it gets more and more powerful and you save yourself time every single week based on what you've actually built. So, another great example of what you can do with their systems and this is a really good example as well. So, Joe shared his setup where he's done two consecutive days of hands-free automation. So, he said, "This morning I woke up, asked Codeex, did Hermes run my content?" And it was a total success. Right? So basically what he's done is he's linked Hermes and Codeex and Hermes can do automated tasks like for example SEO for you which is what he's doing and then it can run on a schedule which means for example you can say to Hermes okay every day 6 a.m. create content to my website and it will just roll automatically and so that's really powerful and uh it's amazing to see like people like Joe building this sort of stuff and getting results with it. So if you're thinking, okay, what can I automate or what's an actual use case? You can see a fantastic use case right here. Um, and he said, you know, this is something interesting to know about Asian operating systems. He said it's awesome. Like literally, he wakes up in the morning, everything's ready to set up. It's truly been one of the most rewarding things he's done so far. This is number one inspiring to see, but number two well done on sharing and also setting up like such a powerful automation and this is a really good post by North as well. So basically North was saying, you know, people who get the best results with AI aren't the ones who have the best problems. They're the ones who set up the systems. And that's the way that I look at it as well. It's like it's not about it's not about prompts. It's not about models. It's not even about Fable 5 or whatever comes out next. What it's really about is having great systems that automate the stuff you do dayto-day. And when you think about things like that, you become an architect and an engineer, not just a prompter. And that's a huge level up. So this is another question. You know, do you think people beginning their journey into tech should focus on learning how to build AI tools, learning how to sell them, learning one niche deeply, or just posting content and building proof? So I think for me personally if you actually want to get clients and results posting content and documenting what you do with AI is really important because it will help you get more clients and also build your skill set. So if you're actually building stuff and sharing it publicly that really showcases your skills and helps you attract more clients. And then I would say whatever you learn on that journey and whatever you build which covers part A as well that will help you to generate more leads. So, I think it's all built into one, which is you build with AI, you showcase what you build with AI, and that helps you get more clients along the way. Another cool case study from Scott. So Scott basically went from zero to two agents in 60 days without being able to code. I can't code as well. And again, I didn't have experience before I got into this stuff, but we still built amazing systems like this. So you can see he says, you know, I run two Hermes agents on desktop, one on VPS. None of this happens without the agent OS framework Julian built. I didn't just learn how to use Hermes. I learned how to think about architecture, the memory galaxy, the orchestrating passions as well. So massive unlock, fantastic to see getting results with this, Scott, and well done for just taking action. Sounds like you've leveled up massively. I think this is proof as well that you don't need to be technical to be able to use AI agents. You can build something like this just like I have. And anyone watching this can build this sort of stuff. It's just about learning how to build the systems. And something like this you can learn to build in like, you know, 30 minutes or a day. Building the music agent literally took me about 10 to 15 minutes. Building out the video agent took me a couple of hours, right? Um but either way, you can take advantage of all these systems because we've got it ready made and built inside the air offer boarding. This is a great question. So Nacho was asking about LinkedIn posts, you know, been trying out for days to to try and automate LinkedIn. What's the solution? How can you get it fixed, etc. Right? For me personally and for my team, we don't automate the actual posting. So we can easily automate the content itself. That's no issue. But for the posting, I don't think the platforms actually like it. So for example like LinkedIn I don't think they they like you to do that. And then number two it also creates more accountability if you have you or your team post the content yourself right because for example if you have to click post yourself then you're accountable to making sure it's actually good. Whereas for example you have an API that's posting for you. You kind of set it and forget it and you don't quality control anything. So for me and my team I always recommend that you can use AI to help you create the post and that's pretty easy. But the main point here is you want to post it yourself and the way that we can quickly like come up with social media content. This is pretty powerful. So if we have a look at the Hermes Oracle built-in the agent operating system, we can for example take a look at this post. We can click on draft a post for this. We can open up the original post as you can see here and then we can quote it like so. And now we've got the post being drafted by the oracle and we can quote the content itself, right? And so for example, if we take a look at this post, if we want to comment on this, we can draft a post with Hermes Oracle, get the social media post and quickly post about it as you can see right here. So as an example, we can open it up, we can quote it, just quality control it before you post it, and then you can use this system as you can see right here. Um, and also if you want to, for example, create content automatically, then I'd recommend just having a blog because a blog can easily create content that's personalized to you has a great system built in. So, for example, as soon as OpenAI's jalapeno chip got announced, we created a blog post about it. It's all personalized and tailored to me with my CTAs and everything else. Uh, it's indexed across three different websites and that's much easier to automate and also it's actually useful content. So Donald is asking has anyone moved the contents of one open claw instance to another instance to keep working without losing anything you've created. So there's a couple of options for this. Number one is you can just have the same instance on multiple different platforms. So for example, if you wanted to access openclaw that's set up on your main computer, then you could use tail scale to connect from another computer and that way you can just have access to the same open claw. You don't need to create multiple versions. You can also create a backup of your existing open claw and then with that backup file you can restore it on the next VPS you set up openclaw with there's a full guide on how to do this but basically you just look at your existing open core setup type openclaw backup create inside your terminal and then you can export that and set it up with your next instance if you need duplicate versions. a lot of cool people inside here as well like people setting up uh local meetups and things like that. The great thing is like inside the community we have a map so you can meet people locally who are building near you and then you can connect with them organize meetups that sort of thing. It's a great way to connect with new people and I think a rising tide lifts all boats. So, if we all learn and grow together, we all raise each other up and level each other up, which is the best thing about a community. By the way, the agent OS does work on Windows as well and on a VPS. So, we got another question from Jim here, which is, you know, I'm brand new here. I've got two VPS accounts with super and NA10 and both MCP connected to Claude CLA for marketing purposes. Blown away with how everything is coming together dayto-day. And then his question is, why would you need a Mac Mini with open core and Hermes? And I 100% agree, right? I don't actually don't think you need a Mac Mini at all with Open Claw or Hermes. Hermes is so light as a setup. And you can also use the blank slate option if you want it even lighter, which means that you could run it on a mobile device. You could run it on a Raspberry Pi. Like you don't need a Mac Mini. I think that's just kind of a marketing myth that came out when OpenClaw was trending, but the truth is you don't need a Mac Mini. And the other thing to note here is a Mac Mini and even a Mac Studio not very good for local models if you actually want a Frontier level or decent setup. From everything I've tested, local models on a Mac Mini or a Mac Studio not particularly good. And I've got a Mac Mini. I have a Mac Studio. So my point here is like if you just want to use Openore or Hermes, stick with something super lightweight. Even the agent operating system that we have, this system, I've run it on like a a MacBook Air, like an old MacBook Air, because it's just like a super lightweight dashboard, like mission control for all your agents. You don't need anything powerful to run it. It's designed to be lightweight and easy to use. Felix joined the group as well. He's using Claude codes. Got started about 3 months ago. Built some voice agents, WhatsApp, etc. This is pretty cool. One thing that I'd recommend actually is take a look at Hermes Jarvis. So Hermes Javvis here we can voice activate it we can voice control it. If I say to Hermes Jarvis open up Google it will open up for me and it's voice activated and it has browser use and computer use as well. Plus it can use the agentic features if we switch that on inside this section. So you can see here we can switch between agent and auto inside the agent operating system. And basically this is a real time voice activated version of Hermes agent that's super powerful. So I'd recommend checking that out. It's a lot of fun to build even if you build it from scratch or if you use our setup inside the agent operating system. That's something that's absolutely blown my mind and it's become like the first time I've truly enjoyed using a voice agent with AI that actually has practical use cases. Wes is asking about paperclipip. Now, if you're watching this and you're not sure what paperclip is or what it is, etc. So, this is basically a way of having your AI agents work together in a team as an organization. So, for example, we have a team of agents here. We can see what they've created and built. We can see all of their builds and preview them over here. And then we can also manage, for example, our costs and we have a full dashboard. We also have an inbox from our AI agents. Basically, this is an open- source paperclip project where we can have teams of agents working together. Now, if we look at where's his question, he said, how do you configure it? So he has his point in the open router but reading some point to Hermes and he did a few agents using a different LLM provider but cut it down to API usage. Yeah. So you can actually use Hermes agent with free APIs or an oar like Grock and then you can have your agents running for free as a team. You don't need to worry so much about tokens. So that's a good option. Some free APIs include for example alpha gro and also if you use noose portal login with Hermes agent you can also use step 3.7 flash and neatron 3 ultra with hermes agent. So those are some good options for using Hermes. Hermes has a specific adapter for paperclipip that's open source and we've already built it into the agent operating system as well. Sasha was asking like can you plug in codeex and openai for the agent operating system and the truth here you absolutely can right so we have a full section for codeex inside the agent operating system as you can see right here we've actually segmented it so you've got chat and you have goal mode for codecs and you can also see everything that you've built inside the sessions and the workspace section as well which is pretty cool. So you can for me I personally prefer to use claude that's why I talk a lot about claude with the agent operating system but with built-in codecs for everyone who prefers to use codeex and then when it comes to using GPT or openi inside the system you could also plug in GPT5 into openclaw or Hermes as well as an o provider which means if you're already using codeex on subscription you could plug that into the Hermes section as well and that way you You can use the codec cla inside codeex without paying for an API and you can also use Hermes without paying for an API and you get two in one if you plug this into the agent operating system but yeah all the functionality is already Yeah. Sean said he just joined and he's already started setting up the agent operating system but doesn't have any of the agents installed like for example Hermes or DeepSeek. Are there any procedures to follow? So inside the installation for the agent OS which is a zip file that gets updated daily, you basically have everything there that you need to get this set up and it can set up any of the new agents for you. So for example, if you don't have Hermes, but you want to plug that into the agent OS, it's already included in the instructions and it's pretty easy to do from there. and we have answered every single question inside the community. By the way, if you want to be able to ask me questions like this and I quick video tutorials for you, answer your questions, whole community helps you as well and you can just get the whole Asian operating system from us. You can get that inside the AI profit boarding link in the comments description or go to the AI profitboard.com. You can see all the cool stuff you get with it. So you get the memory as you can see right here. You get the camb boards, notebook plugged in. We've got all these different tools and workflows I personally use and build with. And then we also for example have like Sakana Fugu built in. We have fusion builtin. We have all sorts of amazing things inside here. Right now, what I do basically is like I answer these questions personally and so does the whole community every day. So, the great thing about this is that you can ask questions and you get help from me plus the whole community. There's always people online. So, there's 112 people online right now ready to help you inside the classroom. You can get access to all of our best trainings. You can go from beginner to expert if you're a complete beginner over here with this course. And then also you can get access to the agent operating system which we update daily. You get the video tutorial, the guide on how to set up and a zip file for installing it. And also as we new add new features in there, you get access to those too. Whenever something new or useful drops, we add it as a guide inside the classroom. You can jump on weekly coaching calls, share your screen, ask questions in real time, etc. And inside the map, you can actually meet people locally who are building with AI agents just like you as well. And that's all available inside the profit boarding link in the comments description or just go to the apiprof.com. Thanks for watching. Thanks, Sheena. Yeah, I did share yours. Really good. Keep sharing it. Keep showing us what you build each day. Excited for you on your clients as well. It really is. It is. It's a bit of both. Bit of both. Nice. Hey, Jen. 11 Labs Higsfield in the video section. That looked really cool to be fair. Need to implement Oracle now. Yeah, that's it. I think it's a really powerful way to get more traffic and also uh to automate SEO with this Uh no, I mean this you might say that like you need you know you might think you need loads of support, technical help, etc. But actually if you check out our testimonials like for example Rick Rick set up his agent operating system in like 30 minutes and Mateo is absolutely loving it as well. You can see Joseph created his like loads of people getting awesome results with this. You don't need loads of technical support or anything like that. Trying to build out an agentic OS and whole agentic departments for small uh businesses. That's pretty cool. That's a cool idea. That being said, should I get a VPS or a local setup? Well, I mean, if you want to give access to your clients, I would there's two options there. So number one is you could set it up for them on their local machine. But if you're not physically with them, then what you could do is just give them access to a VPS with tail scale or something like that. That could be option. You already know my paperclipip is a beast. I do. I do. Actually I do have a question. I have a site on sale. Can I get Hermes to take control of it and then update it? Yeah, absolutely. So you can get the API, get the access token from for sale and then your agents can deploy for you and that's pretty simple and easy. Yeah, is like agents for operating websites is so much easier, 10 times easier because you don't have to log in. You don't have to worry about like, you know, broken plugins if you go into WordPress or something like that. You don't have to create the content yourself. You don't have to worry about design as well. Like I've wasted so many hours in the past trying to design like a landing page with Elementor or for example trying to organize a WordPress page. It's just a mess, right? Whereas for example when you use this system with agents updating your website whether that's on WordPress or for sale or for example on Netifi it doesn't matter. The main point is that your agents can do everything for you and that saves you so much time. Nicely done. It sounds like you're living the dream, mate. Where's the guide? Oh, there we go. Yeah, definitely recommend that. It will save you so much time. It's pretty easy. And then you can set up like a custom skill for it as well. So, as an example that if we check out the agent OS here, right, if you go to the SEO section, you'll see this section called skill. And this basically talks you through our skill for creating content with SEO and how we plug it into Netify and everything else. And then you can just adjust that for your website as well. It's a pretty powerful system because also if not only can you publish content to your website, but you can also pull in the analytics from your Google search console. Then you can generate the content, then you can deploy it and everything's just ready to go in one place. I think that's a fantastic way to use AI. This is great, but can you make sure that you use your guide to create the guide? Showcase all the stuff you've created though. It looks really good. >> Happy to appreciate it. on this as well. Have you added links to the open source projects we use for the skills? So make sure you use your guide skill to create the guide and also make sure you mention and link to and explain each of the GitHub skills that you use for the design systems. And obviously it should tie naturally to the agent OS Heat. Heat. Heat. Hey. Hey. Hey. Heat. Heat. Heat. Heat. Is he Have you got that set up? Not yet. Maybe in the future. You can get my setup inside the air boardroom if you want everything set up. So, if you go if you want the HOS setup that I've got, you can go to the classroom here, go to new daily updates, and you'll find the full setup with the video tutorial, last update, the zip file, and everything else you need to win with this stuff. So, if you want to get my personal update, um my personal setup that comes out, um you've got it right there. There we go. No better than England. Me too. He's coming soon. Yeah, we're getting Nora link in there soon, I reckon. Thanks so much, Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Yeah, got to love the tunes and you agent, he's doing a good job. What's up with Fable 5? Yeah, if you actually rewind this, you'll see all the updates on it. I got a video tutorial coming out soon as well. So, yeah, it's pretty exciting. Um, it looks like it'll be coming out soon. Coming back soon. What does your agent OS come with setup? Would I have to build it from scratch or is most things already? Yeah, you just like I mean to give you an example here like if you're wondering okay how long does it take to set up? How can you set up etc. Rick set it up in like 30 minutes. So it's pretty fast and easy to set up. Um I personally did it. Took me about I think to get everything ready. It's probably about 30 to 45 minutes. Uh but give yourself a bit of time to set up if you want to. You got the video tutorial here. You got the full seat file here with everything ready to set up and ready to go right there. So yeah, you can go from there. Nope, we've already covered Fable 5 twice mate. Just joined. Thank you so much, Baba. I appreciate you joining. That's awesome. Have a good session, mate. Okay, see you later. Back on it. Let's see how many steps you're on. 8,000 steps. Not bad. Not bad. Could do better. We'll aim for minimum 15. Nice. Yeah, I'm testing out some stuff as well. Oh, heck yeah. Guy on a treadmill. This is exactly what I clicked for. Now we're talking. I'm glad you joined me. My custom OS I've built here only runs Claude CLI. Yeah, that's similar to what we've done. Like we got the claw CLI connected, so we can make any changes we want. Let's see the magic you can see in the treadmill. Heat. Heat. Heat. Heat. Hey, hey hey. Heat. Heat. Hey hey hey. Heat. Heat. Heat. Heat. Heat. Heat. Update your skill and all the changes you've learned from today so that you don't forget this stuff. Hey, Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. N. Heat. Heat. Heat. Heat. I notic that you keep misaligning in your guides, stuff like this. You can see it's not lined in the center. And I think you did this yesterday as well. So you can just update your skill for guides like this so that you avoid doing this. It should always be in the middle. That's obvious. Heat. Heat. When I check this as well, like the explanations for how to avoid bad design don't really make much sense. So, please fix that as well because like when I read that out, it makes no sense. Heat. Heat. Heat. Heat. Heat. N. Heat. Heat. Heat up here. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Hey, heat. Hey, heat. Thanks. This is uh from our music agent inside the agent OS. So we linked Hermes agent directly to Sunno API and then it can just generate music in like you know whatever style we ask it for. So yeah, it's pretty cool. Creates some awesome stuff. Heat. Heat. Heat. Heat. N. Back in a sec. baby. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat up here. Yeah, they're talking about it coming back. Fable five, there's a few signals that show it's coming back. If you rewind this, Nice heat. Heat. Heat. Heat. Heat. Today we're going to be looking at some of the best ways to use cord for design uh using an anti-slot machine. So usually every AI built website looks the same, you know, the same colors, same emojis, identity cards, etc. And this is typically what your typical AI generated website looks like. All the landing page builders look exact like this. You always use this gradient color. They love a bit of purple. And you can see an example of what it looks like as a scenario. I'm going to show you what it looks like and how to get better results as you can see. So we can see the before and after based on the design principles that you can teach Claude in like one single click. And these are using open-source design skills. And you can also deal with the skills yourself if you don't want to use anything external as well. Just to put this in perspective, you can see for example here we have a full website created with much better designs using this whole system. And so the way that we generate the images, the frameworks, everything else, it's all using this similar system that we built. So that you can get better responses and better website designs using Claude because Claude or any sort of model, even Fable 5 when it was out was pretty bad at UI. And so this is a powerful way to just wire it in with new skills so that you can get better results. And so if you look at the old way, you know, the you have the AI default, this is what it looks like. Usually you just have like three cards. You usually have like a CTA super generic copy on the page and then like the typical sort of three cards here on on the web design page. With this system, you plug in five design skills. You can just take one of them. You don't have to use all of them if you don't want to. And then for example, you actually have much nicer designs and you'll get better pages that actually look different. So for example, we look at each one of these pages that we've designed here. They all look and feel different. They're all using a totally different vibe. And that's because we've given the new system that I'm going to show you today. So how does it work? Well, what we've actually done is we found some of the best front end design and UI skills that are open source on the web. So for example, like this one. This is a skill for front end design from Claude. We also have this one as well. This is uh from Chad Cen UI. And basically these are different skills that you can install and just you can literally just take the GitHub and you can plug this into Hermes or whatever design tool you prefer to use. And then it shows the design skills to avoid and the design skills actually. So this is another one as well. This is called UI UX program skill. You see it's got 96,000 GitHub stars. Now, one thing to note here is like if you're installing the skills externally, for example, open source ones, you do have to be careful, right? Because anyone can create these skills. Um, so one thing you can actually do if you want to play it safe is you can actually rewrite the skills or you can look through the skill MD or you can also get to look through the skill MD as well and then from there you can create your own version of it. So if you're like I don't want to install these skills separately, no problem. you can just make red versions in the freeways I've just so these are the skills that we've added. You don't have to use them all. You could just use one of them and you'll still get better examples. Now, for example, we look at this one. This actually has 161 industry pallets, which means totally different styles, totally different fonts, and it's matched to the type of product that you're creating a website for. So, depending on the product type, your page will look different automatically. and then you don't need to worry about, you know, all your websites and pages looking exactly the same. And I think that's a great way to sort of look at this and improve this. Then you've also got, for example, the front end UI UX skill. This is pretty cool as well. This is from skills.sh. And this sort of stuff you can install from one click as you can see here. And then you can give your agents like cord or whatever you're using it with a role. So for example, here we have ro designer and developer setup. It has a mission, has work principles, and is basically set up to make sure that you get better design principles from your AI agents. And again, the whole point here is you avoid that AI sloth look that nobody wants on their websites, right? Otherwise, we're just going to have a generation of websites that just all look the same from the same era. Now, you might say, am I actually using this? So all the guides that we create, for example, an Agent OS guide, we use similar sort of principles to just create differentiated guides that don't look so bad. And this this is like a custom skill that we created. And so like every time I want to generate a new guide for any of my tutorials, well, it's pretty easy to just say, hey, use this guide. And then it will create, for example, a nice image. It uses the the fonts. It's got this nice CTA at the top. It's mobile ready and responsive. It comes up with custom frameworks depending on each tutorial I'm talking about. Adds in the sources as well if we're talking about anything external. And then it has like nice quote blocks and nice little blocks like this. It talks about our frameworks and it comes up with custom framework. So everything on this page is created with AI and even plugs in some of our testimonials. And so you can create these custom design skills for each of your websites and then the page itself will look way nicer, right? like how much nicer does this look versus, for example, if we open up those previous pages? This is what it could look like if we just didn't quality control anything. This is what it could look like. This is what it looks nice uh looks like right now. And it just looks 10 times nicer, right? Um I'll show you another example as well. So if we have a look We've created this across multiple different websites and these are just some examples of what you can build. Now the other thing to note here is like you basically have the same copy on each page but you have different words right. So you can see for example depending on the the product that we're creating page for it's going to look totally different and it can have these nice like animations as well. As you can see here it's got some animated sections here. I mean AI is amazing. It understands code better than most humans. But when you're building these pages out, you want to make sure it's just briefed properly to actually use them directly. So what are the moves that make this stuff look better? So number one, you've got starting from everything, right? So build the page around what you're actually using, not a blank template. So when you're setting this up, you want to make sure you don't have a blank template. And the problem with a lot of AI websites is that they actually ignore the subject that you're talking about. So if you're talking about for example AI, well you want like an AI or technology to the website and you can do that with this system here. Whereas for example using like port stra the box it will ignore the subject and just grab grab the same generic layout every single time. Next up for example it has a design principle to make one thing bold. So it just picks one element to make big and memorable. That could be like a strong headline, a striking image, a single animation and then it keeps everything else calm around it. So if you look for example normal AI sloth websites, they just shout everything at once. So nothing stands out. Whereas with this one bold thing, there's more than like 10 busy things at once. Then you have a font with character. So you can choose a font that actually has a personality and purpose with it. I'm going to show you some examples here. So if we look at this website for example, like the fonts are pretty nice. The colors are nice on it as well. And it doesn't use like a generic design. It's like it's themed around the topic as you can see. And then it will just use one color and use it lightly. So we pick one strong color. We use it sparingly on a calm background. So it skips like the traditional colors the AI always reaches for which 99% of the times like the purple gradient and the cream the terra cotta like those are the two colors that usually with AI those are the design colors that it's going to go for as well. And then it gives it through breathe. So this is another design principle that you teach the AI to leave real space around things instead of cramming them all together and use thin lines for example instead of a drop shadow in every single box. Basically you just want to look at all the mistakes that AI typically makes with web design and then switch it up so that you avoid those right? It's like inversing the problems. It's like okay if we wanted to make a really bad website that just looks like generic AI what would you do? Well, here are the things and then you just avoid each one of them and add a bit of movement on the page, right? A bit of animation. So, for example, if we have a look at this, we've got a nice little framework here and it's pretty good for like teaching people new stuff as well. So, you can have these nice diagrams that are animated and use that as well. So if we have a look at the six examples, sorry, seven, you can see the original versus after. The original again, this is what it would look like. Even like the copy is pretty bad. And this is the version after, right? Just looks 10 times nicer, doesn't it? And also, it's unique to the style and the topic. So that's what you want as well. Here's another one. And here's what it would look like after. This one as well. You can see the animations here. It's not using that generic setup. Just looks a bit different. Obviously, this would normally go in full screen, but it's because we're switching between them. And this is like the generic slot website. Another one here. This is kind of like minimal style for a personal portfolio page. Whereas this would be the original. And straight away like as soon as you see emojis outside the title, you know this is generic. Then we have this version and it kind of has these drop downs as well. And finally this version. This is based on my skill. I think this actually looks the nicest out of all of them. Uh simply because when you go to agent OS.guide guide is designed to look like beautiful and it's got a certain feel to the page as well. As you can see, it even uses screenshots from the stuff that we're talking about just to make it look different and original. So really when you're using AI straight out the box for web design, it just makes everything look the same. So this would be the default, the same purple gradient across everything. This is with the new engine that we've talked about today. Everything looks different. And like if you look at the old way, it takes the same amount of time. The only difference is the principles it's using with to create the stuff. So in the old way, you'd have like one purple gradient hero with everything centered. The new ways that you anchor the colors to the subject's real world. In the old way, you'd have emojis in the headings to fake energy. In the new way, you'd have a distinct palette and font pairing per project. With the old way, you'd have like three identical cards with the default font. With the new way, you'd have one signature element, one accent, one real white space. And so that's basically how it works. Also, drop shadows you have to be aware of. And the other thing you will notice is like with the old way, let's say for example, you got a coffee brand and a SAS website. They would basically look the same if you don't apply these principles. Whereas with the new way, every site looks like a different studio made it. And the result is people genuinely can't tell it was AI. Whereas the old way is like, you know, it literally screams AI. I will say this is not perfect. So it's not gonna, you know, people use AI a lot. They'll still probably know that it's used with AI, but it's not going to look like your generic sort of stuff. And some people say, well, I'm not a designer. But you don't have to be, right? The skill is the designer. So you install it once the agent follows the rules on every build. And the other cool thing about this is like if you can't use Figma, which I personally can't, well, you can basically get better designs using this whole system as I've talked about. Also, Claw Design has just dropped recently. You got open design tool which are pretty good for stuff like this. So you might be wondering okay where does skills live? So we talked about it front end design shaden shi uh UIUX pro max front end UIUX and premium front end UI. You can plug those into your agent operating system. They can go into claw open claw or Hermes whatever you build websites for and then every site it builds comes out designed automatically and that could be for landing pages for apps for dashboards etc. So if we have a look inside the agent operating system here, you can see that we can plug this into claude or we could add this into Hermes or into openclaw or whatever we build with here. We could even for example teach skills to our other agents and then use that to make different looking stuff as you can see right here. Now if you want to get the full system, you know, you want this wired into an agent operating system, these design skills are just one piece. But the real power is when they live inside an agent operating system that connects Claude, Open Claw, Hermes into one dashboard with shared memory. So the agent designing your page already knows your brand, your business, and your goals. Right? So if you want the full agent operating system with a 30-day roadmap design, uh four coaching calls a week and daily tutorials, the full prompt library, and a member map to find people near you, you can get that inside the AI profit boardroom link in the comments description or just go to the Aprofit.com. Now some people going to say well good design needs a designer and I'm not one. The skill is a designer. So you install the rules once and the agent follows them on every build. You're not designing your direct and people who have never opened Figma are shipping sites that look handmade. Other people say well AI can't do real design. It always comes out generic. That's the wrong belief really because the right belief is it comes out generic if you have no constraints. If you give it the skills I've talked about and the design principles, you get websites that all look totally different as you've seen today. And then other people say, well, design can wait. Good enough ships, but the look is the first impression. So, a designed page earns attention and a bad page gets closed in the second. So, it's the same prompt, same amount of time invested. It's just that with one, you get better results and you get more attention. So, what did you learn today? You learned how to stop making generic websites with AI. You learned how to make sure every website looks different. You learned how to do this without needing a designer. You learned how to get those free open source skills. You learned how to build it into an agent operating system and also how to make sure that people can't tell you your content is AI. Now, if you want to get my full system on this, you can get it inside the AI profit boardroom. We have the full agent operating system video tutorial, the last update date, the zip file to install it. We had new daily updates based on what's actually working. And if you want the skill personally from me for the agent OS guide section, I'll add that in there too. And then also inside the calendar, you can jump on weekly coaching calls, get help and support real time. Inside the map, you can meet people in your local area who are building with AI agents like you. And if you're a complete beginner, you can go from beginner to expert with AI automation using this 6E masterclass right here. So hope to see you inside there. Link in the comments description or just go to the AI profitab.com. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. Heat. N. Heat. Heat. That's the agent operating system inside the aircraft boarding.com if you want to get access to it. So that's what we're using side by side here. There's a brand new version of GPT 5.5 today, which is GPT 5.5 instant. Um, this has dropped. I'm going to show you what our tests show about it and how it's worked so far. If you've never used this before, basically this straight from open just dropped a few hours ago. Uh so the model should be better at understanding the intent behind your question and adapting to it. Also handles complex constraints more reliably. So you can basically give it a few rules and it keeps up right. Um and shopping and local recommendations also got more useful too. It's rolling out today. Today it's only for subscribers but tomorrow it will actually roll out to free members as well. Bear in mind you can use this with Hermes agent or any sort of agent they actually use. So we've already been testing it. We've created a separate profile for GPT 5.5 instant and it is a lot faster and you can just log in with o. So if you already have a subscription to chat GPT, you can log in with a war into Hermes. You can also use it as the image generation model as well. Okay. Now if you're wondering, okay, how much faster is it actually? Let me show you side by side. Uh I actually just tested out a second ago comparing GPT 5.5 instant which you can see on the agent profile right here inside the agent operating system. and we've compared that versus the default. So this is like the normal model we would use and this is GBT 5.5 instant. So what we can do is we can actually see side by side which one performs the fastest and which one replies quicker. So Let's actually see if they're faster side by side. So, we've got a Grock build into Hermes over here and then we have GT5.5 instant over here. So, we're going to say, okay, tell me a joke and we'll plug those into both of them at the same time. So, this is GT5.5 instant. This is Gro build. Let's compare them side by side. See how they perform. All right. And you can see the countdown here. So, this replied pretty much instantly. As you can see, this one is a lot slower. So, that took about 9 seconds. This one took about 3 seconds to reply, which I think if you're using Hermes or any sort of agent day-to-day, it does genuinely make a big difference and it helps as well. And then also, if we look at the outputs here side by side. So, tell me a joke. Why did the developer uh run out? Because he used up all his cash. That's from Grock Build. Let's have a look at GPT 5.5 instant. Why did the Scarecrow win an award? Because he was outstanding in his field. Both pretty bad jokes, but the point is this is a lot faster and you can just use your existing subscription with it as well. So, it's designed to be like intuitive, smart, and very fun to chat with. It's rolling out to pro first and plus and free users tomorrow. They did an actual example screenshot here. So, they just typed in goat birthday and you can see this like if you're talking about the goat, then today is Lionel Messi's birthday. And then it adds some details here. uh it says happy birthday to Leo with a goat emoji. So the response is pretty nice on the example right here. It's probably an optimized example, but you know, you get the point. Something to note as well here is like GT5.6 was expected. So Mark, you can see here he posted on X about this that it might be kind of like a stop gap measure to make up for the fact that GP 5.6 is delayed, which is pretty interesting. Rarely do I use a model about thinking as well. I mean that that's a really good point. I think what this model would actually be good for is like if you have if you're using Hermes but you just use it as kind of like a chat that's custom trained to you and you don't really use the agentic task or you don't do anything complex with it, it will probably be great for that. Uh so for example, little back and forths like this that you can just do on your phone or when you're out and about or if you you're trying to delegate task to sub agents, that could be another option too. when it comes to actually like building something. Yeah, obviously you're not going to build like a a full agent operating system with chat cheapy instant. So that's something to bear in mind as well. So why is this important? Well, it means you can probably be a bit lazier and shorter and you understand what you mean. So it understands intent better. You can actually stack a few rules and it will follow them. It's better at shopping and local picks. Bear in mind that Hermes now has access to link on Stripe so it can do shopping for you as well if you weren't aware of that and you should get better answers as well. Plus, it's going to be free for everyone as well. Now, when is GPT 5.6 coming out? A lot of people are saying that it will be released late June, but let's see what happens next. If you want to create a profile with Hermes agent, if you want to do it, okay, like how do you actually run GPT 5.5 inside your own agent? So you can do it like this. We what we did is we said Hermes profile create GP5.5 instant and then you can point it via your chat GP login like this and it will actually use your codeex login details. So that's how we plugged it into GPT 5.5. You might also say it sounds technical to use this, but it's just a couple of commands ready to go. What we might do as well if there's enough interest is we could run it on the benchmarks compared to all our other benchmarks for each of these agents. We run the same test on every single model like for example Fusion. If you're interested in that, let me know and I'll run the benchmarks for GPT 5.5 instant as well. If you want to see how it actually handles like day-to-day tasks like you can see right here and we test everything on Goldiebench. So, if you want to see us add that to the leaderboard, we can do. Just let us know. It's not a big Frontier update. So, let's see. The other thing I would say is like you're probably not going to get a lot out of it on chat GBT directly, but if you're using it for Aentic stuff, I could see some real use cases right there. So, just to recap, GT 5.5 instant got an upgrade. It reads your intent. It handles messy stuff. You can set up rules as well. It's coming out for free as well. GPT5.6 seems to be delayed because there's been no announcement of it and you can run it inside your agent like you can see with Hermes agent. It's a lot faster and designed to have a lot more personality as well. Now, if you want to get my full agent operating system where you can run Claude, you can run Gemini, Chachi, uh Hermes agents, etc. We actually have the Hermes Oracle for the latest news updates and social media content. We have Hermes Jarvis which is a voice activated version of Hermes. And then we also have for example memory galaxy and loads of other features inside this agent operating system for mission control. Feel free to get that inside the AI profit boardroom link in the comments description or go to the aprofit bomb.com. This is my community one goal which is to save time and help you scale with AI automation. Now if you want the agent OS you can get that inside the classroom and we drop new daily Hermes tutorials like you've seen today inside here. We also have the agent OS. You can see it's updated daily and you get the zip file to install that. You can also ask questions inside the community and I personally answer them with video tutorials every day. And inside the calendar, you can jump a weekly coaching course as well. Inside the map, you can meet people in your local area who are building with stuff like chat GPT and Hermes. And that's all inside the air profitable boardroom link in the comments description or just go to the aprof.com. Thanks for watching. Let's see what we got here. Thank you very much. It's good stuff. It's good stuff. I need to get some more tunes made. It's still unavailable right now, but it's been released in Claude Code on the strings and then it's also being updated on Amazon Bedrock. So, it should be coming out soon. Let's see. GLM. Yeah, I would say they're both better models than GPT 5.5 in for sure. All right, I think that's all the questions today. We've answered every single comment, every single question, every single tutorial, all the latest updates. Thanks for watching, peeps. I'll see you on the next one. Cheers. Bye-bye.