[00:02] I gave the most advanced Claude model, Fable 5, an hour to start and create a full faceless monetized YouTube business. And the results shocked me. And as proof that this can work, check out this channel right here called Doggy
[00:15] this is evidence this can work incredibly good. So, real quick before we run this test and I show you exactly how to set this up correctly and keep your channel safe, Claude Fable 5 is the most advanced AI model that Anthropic,
[00:29] the company that made Claude, has ever made. And it's really good for super long tasks where it can stay consistent doing actions over a very long period of time. And that's what separates it from the other clawed models. And it's so
[00:43] powerful that the US government temporarily banned Fable 5. and it was just recently made available back to the public. So, we are really going to put it can make us and I'm going to see if this is genuinely useful to you and I
[00:56] with running our faceless YouTube channel businesses. Now, I do want to give a disclaimer that you should not do exactly what I'm doing in this video. I'm essentially going to give Fable 5 full permission to create a YouTube
[01:08] channel from start to finish and I'm going to let it do the whole thing. But in the real world, in practice, I do not recommend anybody does this. You want to quality check every single video before you put it out. You obviously want to
[01:21] research to see if a channel name is taken before you use that name. And also double check it on the WIPO, global brand name database to see if there's a company registered under that name. And you want to have some human input in all
[01:33] of your videos, whether that's the ideation and writing parts of the script of the gaps. That's completely fine. And I've done a video here on YouTube on how still using Claude, but I'm doing this video right here as a proof of concept.
[01:49] for keeping your channel safe is finding video ideas yourself, writing scripts checking them and making minor human input, and making sure that the overall vibe of your channel isn't one
[02:03] that is super spammy and making sure that it actually gives value. That'll keep you safe in the algorithm for the most part. And so even if Claude does 80% for you what I show you in this video, for example, if you hate editing
[02:15] videos, Claude can actually do that for you. If you hate video ideation or do with Claude that will save you an enormous amount of time. And so here's you're going to need to download Claude code onto your computer. Now, you don't
[02:29] have to know anything about coding in order to do this. I don't know anything you're going to install what is called an MCP. This is basically something that connects to Claude to give it capabilities that it otherwise doesn't
[02:42] have. Claude on its own cannot generate pictures or videos. And that's where the sponsor of this video comes in called Higsfield. You can install the Higsfield MCP by going to this page, going into your Claude settings, going to
[02:54] connectors, and adding the information that Higsfield gives you. Higsfield will account with Claude which will then give it the capabilities of generating pictures and videos which are going to be used then in your actual videos as
[03:07] well as the voice over and everything else. And for the purpose of this test in the connector settings, I am giving Higsfield permission to every single thing that it asks for without asking me first. Now, this may be a bad idea
[03:19] because it can burn a lot of credits and maybe produce something that you're not happy with or this is going to turn out absolutely incredible, but we're going to see. But in general, you may want it to ask you before it generates things
[03:31] the right direction and also making sure there's some human input and making sure your videos are valuable, but we're not going to do that in this video. And the niche. And I actually got my niche here from Tube Magic, which is a software
[03:44] called the niche explorer where I always add new niches that I find on YouTube. And I found this dog channel. It's got millions and millions of views on it. old. So it tells me that better quality new versions of these videos are likely
[03:59] is a very evergreen niche. And this is an example channel in that niche called Doggy Digs. And so now here is the master prompt that we are going to give Claude to let it do this. I'll have the prompt in the description for you as
[04:12] well, but my recommendation would only be to use parts of this prompt. Don't doing in this video. Just use the parts that will benefit you the most. And then what we're going to do is open Claude Code and then paste in this prompt right
[04:25] pasted in, the only thing that you need to edit on it is this box right here. is already working. So, all you have to do is paste a link to the channel that you want to model. As you can see, I pasted the doggy digs YouTube channel
[04:40] URL right there. And then you can also set things like the video language, how and other things. And basically what this prompt is going to do is to create a full YouTube channel for us along with video ideiation. And it's even going to
[04:55] make our first video completely for us. And it's going to monetize our channel finding something that we can promote so we can start making money from the doing this, it's going to research competitors and it's going to use all of
[05:10] the best practices that I recommend for writing YouTube scripts and keeping attention, like changing up the visuals at least every 3 seconds on the videos that will actually keep people's attention. It's going to make the
[05:23] thumbnails. It's going to even give us posting packages for these videos. And it's going to do all of it from a single prompt. At least that's what I hope it results are, whether this is worth it, how much that this actually costs me.
[05:36] much it cost me towards the end of this video as well. And then what we're going to do is go down here and select Fable 5 from the model list right here. And I'm going to actually set this to max. Now this again may not be a smart idea. By
[05:50] recommend setting the effort to high or extra high, but due to the fact that we just want to see what is can the best claw model do for creating a full YouTube channel from scratch. Will it be
[06:02] comparable to my channels or something that would actually pass the test as for recommend to other people or are humans still needed? We're going to see in this video right here. And again, a huge thanks to Higsfield for sponsoring this
[06:16] Higsfield yourself, you can do so down at the link in the description. And now, once we've set this to Claude Fable 5 and everything is good, we're going to simply click enter and enter this prompt and let Claude start doing the work for
[06:30] the background. And one of the most annoying things that I'm noticing right off the bat is the fact that I keep having to click yes, allow Claude to do this, and allow Claude to do this over and over and over again. And it's
[06:43] this and I found this file, which is basically a permissions file that you can paste into your Claude project folder with Claude code, and it will basically bypass all of the permissions. So, it'll just keep working by itself
[06:57] and over. and I don't know how to do that. So, I literally just screenshotted Claude, "Hey, can you do this in the folder and it added it itself and that the real world, in practice, you're probably going to want to keep those
[07:13] permissions on just to make sure you don't all of a sudden spend $100 in credits when you weren't expecting to. And another thing I also put in the mega prompt was to download anything required at the beginning of this process. That
[07:26] install stuff. It might just ask you, for example, to install FFmpeg or PowerShell, for instance. But that should be done in the beginning. And you essentially do it for you. And another thing to note as well is make sure you
[07:42] click this little plus icon right here and then click connectors to ensure that you have Higsfield right here turned on. And so now, even though we asked Claude to take around an hour to do this, this actually ended up taking almost 3 hours
[07:55] than that. But a lot of it was taken up with me clicking allow and allow over and over again. So keep that into consideration. But it did take a bit longer. And a big part of that is using the maximum amount of thinking and
[08:08] giving it that giant prompt. It's a lot for it to get through. And it can be hard to estimate exactly how long each thing is going to take. So it did run a bit longer and it took longer than usual due to the fact that I'm having Claude
[08:20] basically build out an entire YouTube business. This wasn't just making one video like I've done in a lot of my previous videos. This was making an entire business from scratch. And I'm going to show you that business here in
[08:32] just a second because Claude has finished and we're going to take a look look at the result of what Claude made, here is how much everything cost in terms of Higsfield generating all of the pictures and videos required for all of
[08:46] the thumbnail images, as well as for a full 5-minute YouTube video and the channel avatar and channel art, the Higsfield total was $40.99. Now, claude for this, since we are running Fable 5, which is the most
[09:00] running Fable 5, which is the most expensive model, would have cost $54.80 80 cents if we are paying directly for the API. Now, I'm doing this within the window right now. There's only a couple more days left until Cloud Fable 5 will
[09:13] be API pricing, and this is how much it would cost. But, I'm using this with my Claude plan that I pay $100 per month for. And it did not use my entire session context window at all. And so, there was room to spare. And so, Claude
[09:26] right here. And so, we're going to take us. As you can see, this is the plan that it followed. These are the phases. required tools. And part of that, by the way, was the fact that we told Claude in
[09:41] that master prompt to ideally use cheaper models for generating pictures and videos to make sure that we're not spending, you know, hundreds of dollars per video like it would be if we were using Cance 2.0, for example, which is a
[09:55] thing it did was it created the brand. So, it made the channel identity, making the avatar, the channel art. And then the next phase was monetization, finding something that meets my personal preferences when it comes to promoting
[10:09] affiliate products, which is making at least $100 per sale. And so, supposedly it found something that we can promote to make between $100 and $150 a sale, script template, and it actually scripted out the first video. It
[10:24] produced the full video completely for us by choosing, for example, even a voice to use and trying to make it as good as possible to get as many views as possible. It created 35 different thumbnails and 33 different video ideas
[10:38] with posting packages for each of those ideas. And then it also gave us a cost report as for how much this all cost. Then it even went through the extra step of giving us a 30-day calendar, a next video prompt, a channel trailer, even an
[10:54] upload checklist, and a growth playbook for actually growing this channel. Now, even if we were using the API cost, you might be saying, "Look, you know, $50 is It really isn't when you compare that to, for example, hiring a coach to help
[11:08] you or other things. I mean, this is incredibly cheap when you compare it to that. But the real question is, is it going to give us actual usable stuff? Well, first and foremost, here is the YouTube channel that it made us. It's
[11:22] called Barkcology. And as you can see, it gave us a profile picture, channel art. It's got a bunch of different ideas for the channel already, all set up, and it's following what I recommend with creating channels like this, which is to
[11:34] face of the channel. And that's due to my personal belief that with AI tools progressing exponentially in the near future, it's going to be incredibly easy to make any type of YouTube channel. And so, one of the best hedges against that
[11:50] so, one of the best hedges against that is having a character that you own and that you can control. That way, AI only helps you because you can use it to push out more and more videos and people have a connection with the character. And I
[12:02] character. How can you have a connection with it? Well, look at channels like Zinny Studio, which makes $15,000 a month, having a simple, clearly animated AI avatar be the host of the channel. And we're going to take a look at the
[12:16] first video that it made us in just a second. But first, let's take a look at how it decided to actually monetize this channel. As you can see, it found us a product called the K9 Training Institute, Total Transformation
[12:29] Masterass. And this is an online program that people can buy and you can make over $100 per sale due to the fact that the product price is $297 and you get 35% of that. And if you get just three sales per month for two consecutive
[12:44] months, you get 50%. That way we're making $150 per sale, which is more than possible. And it gave us the link as to where to sign up for this. Now, Claude couldn't actually sign up for it for us due to the fact that it doesn't have,
[12:57] for example, my payout information, my bank account information that it would need to be able to do that, but Claude's telling us here why it fits Barcy's audience perfectly. And it gives us some potential earning math right here. It's
[13:09] taking that channel into consideration due to the fact that that one channel was averaging almost 900,000 views per video. So, every single video would have dollars. But don't expect results like that with running a channel. Some of
[13:23] they're very evergreen, meaning they can keep getting views for years and years in the future. Look at every video you upload as a seed that you're planting. It may not come to harvest immediately, but you're planting those seeds at least
[13:36] so they can come to harvest for you months or even years down the line. And here as well. Probably due to the fact that we're running this on the max thinking power. So, it's just going that extra mile on everything here. And it
[13:49] even gave us an FTC disclosure to put in our description saying that this description contains affiliate links. I may earn a commission at no extra cost to you. That's really essential to also put in your descriptions if you have
[14:01] affiliate links. And speaking of the extra mile, it also made us this channel even ask for a channel trailer. I just asked it to make a channel and do anything else you think will be really good. And it made us a channel trailer,
[14:14] channels. It's not something I even really focus on or recommend. In fact, typically going to want to put an incredibly long video that sells you a lot more money than having something like this. This is basically
[14:29] just a quick 30-se secondond introduction to the channel and stuff that it posts and everything using the animated character and the voice associated with it. And Claude also did an incredibly deep dive into what was
[14:42] working on this channel and the aspects that were working. And it also looked into other channels that are related to this channel and it found what was working on them as well. And then it started finding the patterns in that.
[14:55] And it found outliers on those channels that were getting the most amount of views relative to their channel subscribers. And it was smart enough to know in finding those related channels was to not imitate the trainer channels
[15:08] human expert on camera if you were making dog training videos. These are going to be animated videos. So, we can't have a trainer. So, it was smart enough to know that we do not want to be getting ideas from those channels. And
[15:21] instead, it's focusing more on the psychology of dogs with explainers and curiositybased titles executed with a consistent animated 2D mascot host, which is what we asked it to do. And it found the title patterns had
[15:34] overperformed. For example, the real reason dogs insert any common behavior that dogs do is insert emotion, disgusting, weird, scary. And then you patterns here as well. And then it gave us these 33 different ideas and it
[15:49] ranked them for us on what is likely going to get us the most amount of views. And the number one video idea that it gave us here is 10 everyday habits that secretly hurt your dog's feelings. That's great because it's
[16:01] don't want to hurt their dog's feelings. And you could have a great thumbnail with that as well of something that you always do with it most dogs like maybe petting its stomach or even just petting the dog in general and then saying like
[16:13] get a lot of people to click. And then ideally you would explain in the video that only petting the dog in this wrong way is bad or something like that. Make sure it's truthful obviously and not just clickbait, but that's how I would
[16:26] did it right here and if this is something I would realistically use on a channel. And so it gave us all of these ideas right here. These are incredibly thinking behind them as well with combining ideas that have already
[16:40] worked. And so that's something that I can give my stamp of approval for. And so what I'm really excited for now is the video itself. Is this video going to be any good? And is it something that I can actually get behind? Let's take a
[16:52] look right here. at least three three ways you showed your dog love to. >> So, first things first, off the bat, it's a great hook saying you've probably done these three things already today. So, it really opens a loop of curiosity
[17:05] in people's minds that they're only going to get closed if they continue to watch this entire video. So, that's great. Something that I don't love right off the bat is the fact of how small these captions are. In fact, I would
[17:18] video like this. I honestly don't really think they're necessary, so I'd probably get rid of those. But we were giving Claude some creative freedom to do what it thought was best with this. And it put captions on the videos, but again, I
[17:31] people watch YouTube on mobile devices. I would have just put no captions there. And then if I wanted to regenerate this, I easily can. We already have all of the this generated on my computer. And so therefore, it wouldn't cost me any extra
[17:46] money aside from my claude usage to regenerate this video without captions. But I like the character here. That's cool. Uh the animations are cool as consistent theme throughout this as well. Let me take a look at this here.
[17:59] >> So this is animated actually. As you can see, it's moving. It's not just a static image. If we were using static images, it would have been a lot cheaper. This would probably be a $5 to$10 video, but it's a bit more expensive due to the
[18:12] it will help keep people's attention. Let's see if it listened to me here and if it switches scenes after 3 seconds. >> Registered as stress and science can >> So, as you can see, we're 7 seconds into this right here. And it is a continuous
[18:26] scene, but it switches into another scene midway through here. So, it is let's continue. >> Nearly every owner does it at the door each morning, and it can wreck their whole day.
[18:40] than 3 seconds. Let me take a look at that again right here. >> Nearly every owner does it at the door each morning and it can wreck their >> So, that was a 5-second video right there. There may be some minimum video
[18:53] here where it can't generate a super short video and we may have to ask Claude to manually cut clips themselves and it may be generating 5-second clips it looks like here by default. So, that may be what's going on with the model
[19:06] continue. These 10 come from peer-reviewed studies K9 cognition labs, not internet folklore. Count how many you did this week. Habit one, the hug. >> That's cool. I like this animation. >> Study of 250 photos. 81% of hugged dogs
[19:23] >> Dogs are built. >> And so it's starting with something that >> And so it's starting with something that people do all the time. Um, and make people want to watch more. So that's fantastic.
[19:35] >> A hug deletes their escape plan. That's restraint, not affectional. restraint, not affectional. The tells pinned ears, a turned face, a >> That's good. And I also like that the voiceover is consistent. I was a little
[19:47] some previous Claude models where the voice isn't always perfectly consistent consistent voice, same voice as the channel trailer as well. So, you'll know that all your future videos will also be
[20:00] this voice. But something I don't like about doing all of this in one prompt is might do a slightly different animation style to this and and I can't really edit that for two hours as it's doing everything. But overall, I'm pretty
[20:13] impressed with the video. It's a good video. I wish it listened to me asking to switch it every 3 seconds a little bit more, but the video itself is really here. But what I personally would do is I would look at the animation styles
[20:25] that are actually working on YouTube in this niche and I would do something any with this exact style doing incredibly good. However, I'm sure it could. It's very unique. It gives you a nice brand feel, so to speak. And this
[20:40] could do incredibly great. This is a pretty darn good video. Again, I would ask it again as well and probably follow up and say, "Switch the scenes at least pretty good video. And it's good that we generated one video instead of 33 at the
[20:55] same exact time because if we did that, we would be spending a ton of credits and we'd have to then go back and redo each and every one. So doing one video as a test like this is always a good idea. Remember to write scripts yourself
[21:07] or at least sections of them and have a human review anything that AI does before you upload it to YouTube to make sure that you're actually giving people you take a look at that channel Doggy Digs, you'll notice that all of their
[21:20] videos are purely stock footagebased. Now, that has some real limitations with it due to the fact that if you're purely using stock footage, you can, for example, get a clip of something that you're talking about if it doesn't exist
[21:33] in a stock database. So, for example, if you want somebody hugging a specific breed of dog to show at that moment, and it doesn't exist on a story blocks, for for that. And that's where AI video comes in because you can generate those
[21:48] just a second an alternative workflow you can use to save you tons of time while also not costing you as much credits either. But first of all, I want to go over did this live up to my expectations using Claude Fable 5 at the
[22:05] entire YouTube business from scratch. And honestly, I would say for the most part it did. The things that it didn't live up to my expectations were things that I didn't explicitly ask for in the beginning with the exception of
[22:19] switching scenes every 3 seconds. That is something that I did ask for and it didn't follow exactly, but it still gave me a great video and it gave me a great result. However, realistically speaking, you're probably not going to, you know,
[22:31] generate a full YouTube channel from a single prompt, which is where Fable 5 at the maximum thinking power really shines at. So for most people due to the fact that Claude is now switching to API pricing in just a couple of days for
[22:46] Fable 5, in my opinion it makes a lot more sense just use Opus 4.6 or Opus 4.8 for doing the particular task that you need to do and just do singular tasks
[22:58] for your channel. For example, if you just want to make a video, just have a task of doing that video. If you just want to make channel art, just do a task for that. I wouldn't recommend doing this master prompt and everything and
[23:10] therefore you don't really need Fable 5 for this. Honestly, it's too powerful for this use in my opinion. So, I personally will only be using lower cla models. I really like Opus 4.6 for writing and Opus 4.8 for, for example,
[23:25] assembling a YouTube video or something is more than good enough. And so, yes, I would say I mostly approve everything that Claude did with a few exceptions, but humans are definitely still needed in this process right now in my opinion.
[23:38] creativity, please do it. I'm just showing you how you can hopefully save for doing that. And Higsfield is also an essential tool that I absolutely need out down at the link below. And now, a lot of people have been asking me, I see
[23:52] so many comments asking, "Matt, YouTube doesn't allow these types of videos." They say that, you know, YouTube is banning things that have to do with completely AI generated visuals and other things like that. And it is true,
[24:06] YouTube has banned a lot of channels, but there are some similarities of all those channels that almost nobody's talking about on YouTube. People are saying, "Oh, it was the AI voice that caused it or it was using an AI visual."
[24:19] That's not at all what caused it. And I actually made a video that shows you my entire process of making this into a real business and making as much money on YouTube as possible. And at the end of that video, I go over some 2026
[24:33] updates for doing this. And it also shows you exactly how to keep your channel safe and what's actually causing channels to get banned if they're using AI. And so I highly recommend you check out this video next to me right here.
[24:46] that are getting channels banned and how that's at the end of this video. But at the beginning, I'm going to show you how to make it into a business. Check it out.