---
title: 'I built a fully AI automated GTA 6 YouTube channel. Here are the real results.'
source: 'https://youtube.com/watch?v=PyHyAJ2c3rI'
video_id: 'PyHyAJ2c3rI'
date: 2026-09-09
duration_sec: 1438
channel: 'Silent Snow'
---

# I built a fully AI automated GTA 6 YouTube channel. Here are the real results.

> Source: [I built a fully AI automated GTA 6 YouTube channel. Here are the real results.](https://youtube.com/watch?v=PyHyAJ2c3rI)

## Summary

The creator built a fully automated YouTube content system using Claude, Remotion, and Eleven Labs to launch a GTA 6 news channel. The experiment aimed to test AI's content creation capabilities and reach monetization before the game's release. After one month, the channel 'GTA 6 Hub' generated significant views despite algorithm suppression, offering key lessons in automated content production.

### Key Points

- **Experiment Setup** [00:00] — Creator built an automated system using Claude to research,  script, edit, and publish YouTube content for a new channel named 'GTA 6 Hub', targeting monetization before GTA 6 launch.
- **Content Quality** [01:01] — A popular video with 3.6k views demonstrates the output quality, covering specific details like NPC reactions to weapons in GTA 6, which contrasts with generic content.
- **Pipeline Architecture** [04:14] — Claude used a 'skill' to manage research, script writing, voiceover via ElevenLabs API, editing with Remotion, and quality review. It logged findings to avoid repetition and self-optimized based on video performance.
- **Results: One Month Stats** [10:01] — The channel posted 167 videos (primarily shorts) across YouTube, Instagram, and TikTok, achieving 105k total views. While not enough for monetization, YouTube, YouTube, YouTube outperformed other platforms significantly.
- **Platform Suppression** [11:45] — The creator noticed likely suppression due to accounts being linked via the same device/IP. TikTok collapsed completely, YouTube channel, YouTube. It was the most promising platform.
- **Top Performing Content** [18:08] — Specific, niche topics (e.g., 'T-shirt confirms radio station') outperformed generic titles ('10 things to know'). Claude effectively identified high-performing subjects.
- **Overall Lessons** [20:26] — Key takeaways include: the pipeline works; avoid building accounts on the same network as personal ones; content quality is not the bottleneck; TikTok suppression is structural; YouTube shows the most promise.

## Transcript

Over the last two months, I designed and built a fully automated system using Cloud that researches, scripts, edits, and publishes YouTube content with almost no manual input from me. The goal was simple, to get a brand new channel to real monetization before the biggest game launch in years launches.
This is how I built the system, what happened when I actually turned it loose for a month, and whether it was worth it or not. First and foremost, let's talk a little bit about the actual channel itself. I created a YouTube channel called GTA 6 Hub.
This is supposed to be a channel that was trying to get to monetization ahead of the launch of GTA 6. I think GTA 6 is definitely going to make a lot of millionaires, even from a content perspective. So in an attempt to try and get in on that before the game actually launches,
and in an attempt to test how good Claude actually is at creating YouTube and social media content, GTA 6 Hub was born. But this channel is linked in the description if you guys want to take a look at it, and we're going to deep dive how I actually built this channel and the stats behind this channel.
Let's just take a look at one video really quick so you guys can see the quality of the content that I was getting out of Quad. So first and foremost, let's filter by popular. We can see that one video actually got 3.6k in views. Let's take a look at this one.
In old GTA games, you could carry a rifle around and nobody cared. Rockstar just told Dave Magazine, that's over in GTA 6, their words. Walk down the street with a rifle, and the city will have something to say about it.
NPCs and cops are supposed to react just to seeing a weapon, not only when you use it. Leaked footage already hinted at this. People reacting to Jason just carrying a knife. Tonight's the first real gameplay reveal, so watch what happens the second a gun comes out.
One thing I want to make crystal clear before we start going through this video is that I think that there are two different types of people when it comes to YouTube, or at least two different believers on content and AI's role in content creation.
They're the people that really believe that a lot of it isn't art, and I'm one of those people. I really care about the content that I put out on this channel, and I try and make it the best quality that I can. But it's undeniable that platforms like YouTube and Instagram are monetizable in their own way
and often more lucrative than some daytime jobs that a lot of people have. So, in a sense, social media also has become its own economy, and that's really what this channel is for. The goal of this was monetization, not necessarily enjoyment,
and also trying to see how far I could really push Quad in the production process. So this channel was created on YouTube, Instagram, and TikTok, and again, all three of those are linked below if you guys want to check them out. Without spoiling too much on the stats, let's talk about the build.
So again, the goal was to reach monetization, and I think that's really difficult to do nowadays, especially I believe YouTube actually just raised the requirements even more in an attempt to push back on AI-generated channels.
I think that's really interesting considering what we're going to talk about stats-wise in just a little bit. With how high the requirements really are, no matter which path you take, whether that's short or long form, I think a lot of people in today's day and age, it's borderline impossible to do.
If your goal is monetization on YouTube and making money, you have to have some sort of system and production process. Otherwise, it's borderline impossible. I mean, even like Moist Critical, who seems to talk about pretty much anything
that's on his mind. He has a production process and a pipeline that he follows to be one of the first to talk about things and then also get his videos recorded, edited, and posted in a very timely manner so that he's one of the first to whatever stories he's talking about. And that
applies to a channel in almost every single niche. So that's why I wanted to come out here and try and create a system with Claude that actually creates content for me and can auto-publish it with little input from me. So at first I started with the 10 million shorts views pass, but I
quickly realized that that was very, very difficult to do in 90 days. So 90 days is the measurement window that YouTube is using to determine whether you hit 10 million views or not. So this isn't total views. This is 10 million views within 90 days on Shorts exclusively, which is very,
very difficult to do. The option most people take is 4,000 hours in long form content. And that was something I also very shortly tried. And we'll talk a little bit about that. But this is the kind of goal that we're trying to beat, and we'll take a look at the stats in a little bit on how
close we actually got in just one month. What does the pipeline for production actually look like? First and foremost, everything was built with Claude and on my own PC. Actually, Claude didn't really even build anything. It was more just the skill that I gave it to use the tools I had
installed, which was mostly Remotion and also an Eleven Labs API for voice generation for all the content it created. Basically what I did with Claude was had it create a skill where it's
actually researching, scripting, voiceover using 11 Labs API, and editing using Remotion, then reviewing the video that it creates, and then finally letting me know when it's done. So this whole process was maintained within one skill. That skill was originally created
for short-form content, which was then repurposed to make a couple long-form videos, which again we'll talk about in a minute. Well I think the big learning with this process was being as detailed as possible. Like with the research, I didn't tell it to just research every single day anew. I
actually told it to make sure it's logging things that it finds and dating that so that it's not repeating or finding the same information over and over again and thus creating the same video over and over again. On top of that, it's also actively looking back at the videos it's made
that are in an archive folder to see what has it already talked about and make sure that it doesn't talk about something it's already discussed in the past. And I have to say, that was probably the most impressive part of this entire pipeline for me was the fact that it not only was actually
looking back at what it had already discussed, it was actually creating new ideas, well, new, so to speak, branched ideas is maybe a better way to say it, that built off of what it had already talked about, almost like a real creator would.
It would actually save scripts for each video that it made so when it did go back and review what had already been done it would reread the script and actually even call back the specific lines in future videos So this was a really, really important step and one that I think was the foundation of the entire
process. Scripting was also very important, obviously, and I did start studying kind of on a video-by-video performance basis which types of words or topics seemed to garner the most attention,
And as we noticed that in the data, because it was running, I think, a weekly review, would actually start integrating that more, and ultimately performance kind of started improving slightly. So not only did it do a great job at actually picking the topic,
but then actually reviewing what it had already made for further developing scripts and creating a better channel overall. This over was pretty straightforward. For those of you who have played GTA V, I'm sure that words are familiar to you.
For the purpose of this channel, that's what I decided to go with. The argument for using DJ Kara's likeness in some sense was a retention buff, essentially. The idea was you're hearing a voice that's familiar, kind of like the automated computerized voices that TikTok had for a while, right?
Like we all kind of know that voice. Or Siri, for example. Kind of playing that same idea, but it would have been interesting to try with a different voice for sure. The editing was also obviously instructed, you know, things like subtitles, the subtitle color, subtitle animation.
That was all manually kind of controlled by me in some way. I didn't actually do the editing, but I guided what the font should have been, what the color should have been, where it should have been positioned. Same goes with all the images that are put on.
One thing to note about visual editing, you may be wondering how did it actually get the content to create the videos. All of that was pre-recorded and put into a specific folder, and that directory was then given to Clog within the skill it built to create content,
so that it always knew where to find the video footage that it needed. Every single clip that I had in that folder was labeled with what was actually happening in the footage. so it was able to figure out pretty quickly what it needed to use depending on the situation.
For example, there are some videos where it's talking about cops in GTA 6. But because of that, it actually looked for footage that I had in the database regarding cops or being in a cop chase and used that as the footage for the video.
And finally, it would render out the video with Remotion and do a quality assessment check, basically pulling, I believe it was 5 to 10 frames from each video, making sure all the text looks right, everything's positioned properly. and then once it assumed that it was completed, it would review and publish everything.
And now again, this was all packaged into one skill, and then that skill was packaged to run once a day at 9 in the morning, and it would create three videos every day. Once those videos were created, they would all be output to the same folder.
I would review them, drag them into Google Drive, and from there I had a MetroCool MCP connected to actually find those videos in Google Drive and schedule them on my behalf. So that was the only manual part of this process is after it actually completed editing the videos, I would go back, review them for my quality assessment check, ask to make any changes, and if they're all good, I'd upload them to Google Drive and reply on the times that I wanted them to be scheduled.
Not necessarily a straightforward process. I would definitely say that if you're planning on making some sort of automated content production with Claude or any sort of agentic harness, definitely make sure you have a pretty clear vision of what you're trying to do.
don't expect it to be right the first time because it definitely did take some trial and error to get the skill where I wanted it to be but once it got to the point that I felt was of high enough quality it was able to repeat that process over and over again day after day. So like I said I
had two different skills that were created the first one was the short form pipeline and here was pretty much the basic guidance for that skill everything 9 by 16 here's the time section things like that and then like I said this was actually repurposed into a long form content skill which
did not get too much use and we'll talk about that in just a second here. So the results, how did it actually go? Well here is the stats for August. So again this is only one month of running the channel. Overall I would say not too bad. Definitely nowhere near where we need to be
on the 10 million short views in 90 days. So the big comment being that I think it's very difficult for Quad to make quote-unquote viral content. However I think a lot of that also rests in the
footage that you provide and the guidance you give on the editing style and any edits that you do afterward. And I think that there was something impacting the channels here that we've talked a little bit about in regard to AI influencers, especially those of you that have taken the
course, which you guys can find in the description below. And we'll talk about that in just a second. But overall, not too bad. Total views across all the channels was 105k, 167 videos posted with 165 of those being short, a total of 1,500 likes across three platforms. And so YouTube was by far
the top performer doubling Instagram. And this is why I think it's so interesting that YouTube has been quote-unquote cracking down on AI-generated content and AI-generated channels, because Instagram and TikTok definitely did a better job at suppressing this content. Like, if we look at
TikTok, I was posting pretty much the same videos on the same cadence, and there was a point where TikTok views just like fell off a cliff. And I think part of that is not due to the content quality necessarily, though that's always a variable. I think it really has to do with an
idea that was discussed, which is the fact that social media platforms like YouTube and Instagram specifically in regard to AI influencers are definitely able to track which accounts are connected together or owned by the same person. All three of these accounts for GTA 6 Hub, the
YouTube, Instagram, and TikTok were all made on my personal home device, which is the same device that I use to post for Silent Snow, to post for my personal accounts on the same Wi-Fi. So what a
lot of people have noticed with AI influencers, especially on Instagram, is that if you are running the account from the same device that you use for your personal account and you post on the influencer account on the same Wi you use for your personal account Instagram will often link the two together and actually start recommending the profiles to each other And that essentially a sign that they know that your accounts are connected
and that often suppresses the content from that new account that you've created. So I think that's largely what happened here because I was starting to see that with YouTube where content from GTA 6 Hub was being recommended to my Silence Now channel
even though I haven't subscribed to it. I tried to keep them isolated from each other but I think that's largely what happened. And because of that, I find it just even more interesting that YouTube claims that they keep cracking down on AI content when really that doesn't seem to be the case.
I think I was pretty consistently able to get at least a thousand views on YouTube with shorts. Granted, none of them blew out to like the 20k view mark or anything like that. I think that's, again, partly content suppression, partly content quality.
I just found this really interesting that YouTube did seem to perform the best and often is the most monetizable of the three. If I had to go back and do this process again, I would probably focus on long-form content first,
or I probably would have built the skills together, because I didn't do long-form until after I had already perfected the short-form skill, and by that point, YouTube algorithm had kind of already considered me a more short-driven channel,
and because of that, the YouTube long-form content did not really perform well. So I likely would have gone back and started with long-form first, So that's probably the easier path to monetization. But Shorts still worked. There's just no way that you've reached 10 million views in 90 days,
even with an automated production process, with YouTube suppressing that type of content and then also accounts that are associated together. Here is the Instagram breakdown. We can see some kind of wild performance here.
So this little section here, by the way, was actually kind of a quiet period. That's why it was so low. But even looking through kind of these peaks and valleys, are up and down there's really no consistency and this again is why I think that sure enough
these social media platforms really are cutting down on bot activity bot accounts AI driven accounts and I don't think that's fundamentally a bad thing don't get me wrong guys like I'm not on any particular fence of this conversation I just thought that this was an interesting
experiment obviously like I said GTA 6 is a massive launch and one that's going to make a lot of people a lot of money so why not try and hit two birds with one stone and I wanted to share of my findings with you guys. But I think it's a good thing at the end of the day that YouTube
really is pushing down hard and really all social media on completely AI generated channels. I will say that there's definitely still some substance to the channel that was created. Like the news that it was giving, it was actually pretty good news and things that I hadn't even found on the
internet sometimes in regard to GTA 6. They did a good job of, you know, staying close with the whole leaker situation that was happening and covering that and covering the launch on Netflix and all these kind of more novel, up-to-date news pieces,
it wasn't just regurgitating the same content over and over again. And I think had I given it a more diverse pool of footage to work with, we could have seen potentially better, more personalized results here that really kind of bypassed those restrictions.
I think automated accounts aren't necessarily a bad thing as long as they still have some taste and are providing some value. Someone like me who is really excited for the launch of this game would probably follow an account, even if it's completely AI-generated,
because it's keeping me up to date on the news and it's an automated system that literally is focused on keeping you up to date. It is not waiting for human latency, right? So I would kind of be okay with that, but there are a lot of people obviously who wouldn't be. So again,
not bad performance, but we're just hovering at like one particular peak. There's really no breakout here. And then TikTok again was a very interesting story. It totally just flatlined, Like we were, you know, actually looking really good and really positive.
And come August 8th, it really just completely flatlined. And if we take a look at the channel, you guys can see it's just a sea of zeros here. You know, we talk about getting ghosted because they're not posting enough or the content being regurgitated.
And again, visually, you definitely can see that there's a patternistic approach here. But I mean, we had some videos that actually didn't do too bad, but eventually it kind of just stopped. This is where it starts to become a question of algorithm for me, because at a certain point, there's not much you can do if you're being suppressed.
Like we can see that TikTok did not do well at all, only have four followers. And this, I think, is just a good example of what happens when a social media platform is kind of detecting your bot activity. So if you are planning on doing something like this, I would highly recommend that when you are creating your accounts,
If you're doing it on your PC or the same device that you have your personal accounts, make sure you have a VPN on to kind of separate that connection between the two. Better yet, try and do it on like cellular service or something. Yeah, TikTok complete flop, 38 videos with zero views, largely due to algorithm suppression.
And here is our chart for YouTube. So again, 70,638 views. That's quite a lot, guys. I'll tell you that on the channel that I've been running for over a year, I think we are just now rounding 300K.
So 70K views, even though that's just shorts, in one month, pretty solid for a completely automated channel. On a total of 60 published videos, which includes two long-form videos, and 214 watch hours.
Now, shorts watch hours don't count towards the 4,000 long-form hours needed to hit monetization, and this is why I said I would go back and do long-form first, because if we were able to get 214 from short-form, and if we just refocused that energy on long-form,
let's say we were able to post two long form videos every other day. Each of those got, I don't know, let's say 5,000 views and they were each five minutes long. That's quite a lot of watch hours. You'd probably be able to hit monetization pretty quickly. Like here is one
day on 820 where we got over 7,000 views in one day. And then obviously this low period here, where the pipeline was kind of going through some maintenance Overall I mean most days were over a thousand views if not close to two thousand views But overall yeah solid performance from YouTube which continues to shock me considering all the efforts quote that they been making against AI content on the platform Which videos actually perform the best
and why? That is always the question that you should be asking when you are running a YouTube channel. One of the biggest lessons that you can learn on social media as a creator is if it ain't broke, don't fix it. So what is not broken that we could have capitalized on and what was broken
and why was it broken. So just taking a quick glance at the top performers, the top one was a t-shirt just confirmed GTA 6's first radio station. So very niche, very kind of click on me type of title. GTA 6's trailer hit this exact date three years ago. Again, very kind of click
on me. Here's another one in regard to the pre-sale. And I'm looking at underperformers, very kind of general content, right? GTA 6 trailer three could drop in six days. 10 things you need to know for GTA 6. These are kind of the big titles that are already owned keyword wise by
other channels that have talked about this maybe even like over a year ago now. GTA 6 content niche specifically is very saturated because there are a lot of people who see the value in having a monetized channel come the launch of this game but this can apply to almost any channel right
like let's even think about something like Blender 3D animation. Up three things you need to know when getting into Blender, right? Like that's going to be owned by a channel that already has a lot of kind of velocity in the niche, in the category, whereas something like this is a little
bit more open or specific that maybe a channel that's covering something like this. So very difficult to break into a category like this with topics like this. If you're planning on making an
automated channel, definitely try and do something that's a little bit more specific and niche that maybe other people aren't covering. I think this also just speaks to how good Claude really is at understanding this idea because it was constantly looking at topics that performed
well and which ones didn't and was optimizing itself towards the high-performing videos and topics. So, you know, something like this I never would have found or focused on, right? But it did its own research and actually found this itself. It was probably covered by some other page, but I
know that there were a couple of videos in here that it actually was able to determine itself just based on research and looking at images. So what were the overall lessons? I think we've touched on a lot of them, but let's just tie them out here. The pipeline works. If you are making
automated content with Claude or any sort of agent to Karnas, it does work. It will get views. I think that there's a very specific way to do it. Like I said, avoid creating accounts on the same Wi-Fi and systems that you use for your personal accounts or else you're going to be
kind of blacklisted just as bot activity. I think that was the biggest kicker. Without that, there's a good chance that this content probably would have done a lot better than it did, but it does work. Having Remotion as the video engine, giving it prepackaged video to work with,
giving it a connection for voice generation, all worked together and Claude did a pretty seamless job at putting the videos together. Even when creating the skill out of the gate, it actually kind of knew exactly what I was getting at. The video wasn't great, but even on the first test,
it was able to put a video together with the content that I gave it. So if you are in content production at all, I would definitely highly consider to check out Remotion because I definitely found Claude to be a pretty helpful assistant in all things editing. Content quality did not seem
to ever be the bottleneck. I do think that if the content quality was improved, the content probably could have performed a little bit better, but I don't think it would ever have reached the kind of 20-50k view threshold that I was looking at. I don't think that improving that quality would
have gotten you there. I think a lot of it has to do with YouTube and social media algorithms, and that brings me into number three, that TikTok's collapse is definitely structural, not creative. The same structure that was used to make the videos that were getting views were
used to make the videos that were not getting views at all. A lot of that was TikTok's algorithm pushing down on my channel, not the content itself. Volume cuts both ways. The advantage of having an automated content production system is that you can produce a lot quicker. Now that
usually comes at the sacrifice of some quality, but it is a huge advantage, especially with the requirements that are needed to hit monetization on any platform. But overposting on a very specific subject with like a repetitive format is obviously going to reduce the quantity of performance that
you get out of your content. What I'm essentially saying is if you're making three videos structured the same way about the same topic, two of them are probably going to get somewhat suppressed. Number five is that, you know, to me it seems like YouTube is the platform they double down on.
Now, I'd be careful with this because YouTube is really cracking down supposedly on AI-generated content and platforms. Now, I don't think anyone fully knows what that means. I've heard of channels that are getting completely banned even though they've never used AI. I've heard of
channels that are exclusively using AI that are doing fantastic. So it really seems to be a mess over at YouTube right now when it comes to fully AI generated content. But based on the performance that I had over the last month, which came after the announcements that they made regarding lifting
requirements for the monetization paths, not much has seemed to change. That was it. I just wanted to walk you guys through this process. I hope that this was a little bit enjoyable. I know it's different from the content that I usually make, but something that interested me and I know that
we're talking about all things AI on this channel. So let me know if you guys have any questions or comments based on this. What are your thoughts on automated content? Is it something that you guys are actively doing or considering for your profession? If you're interested in AI influencers,
consider checking out our AI influencer course linked in the description below. You guys have really loved that so far, so thank you to everybody that's bought in. But if you're just looking for some additional resources, feel free to also check out our Patreon linked below. But that's it guys,
I will see you guys in the next one.