Create this NEW Ai NICHE and go VIRAL in 48 hours | Using 100% FREE Tools

0:00Now, in my last video, I showed you guys how to create these new trending Vox style animations, and you guys loved it. The video went crazy viral. It's currently sitting at around 100,000 views. But, I got a bunch of comments from people saying the workflow I used was a little too technical, and asking [music] if there was a way to create these videos on autopilot without making everything so complicated. So, I went back to the [music] drawing board, and I finally found a way to mass-produce Vox style animations on autopilot. And in this course, [music] I'm going to walk you through the exact method I discovered so you can start bulk

0:31creating your own Vox style animations right now. All right. Now, the very first thing you're going to need is an agent to automate the workflow. Now, there are a couple of options in the market. If you're a completely free user, and you don't want to get a subscription, then the best one for you

0:43to get is Anti-Gravity. Anti-Gravity is Google's AI agent. Just download the version of Anti-Gravity that's compatible with your operating system, and then install it like you would any normal software. Another good option is Codex. [music] Codex is OpenAI's agent software. So, if you've got a ChatGPT account, you can use Codex. Just download Codex for your

1:03operating system, and then install it. And then the last one, which is very popular, is Claude Code, which is Anthropic's AI agent. You could use any of these agents. Claude Code seems to be the most beginner-friendly. So, just download the version of Claude Code that's compatible with your operating system, and install it like you would any normal software. All right. Now, for this walkthrough, we're going to use Codex. So, when you download and install Codex, it should look like this. It's pretty much like GPT, but the agentic version. What you've got to do right now is click on the modes. Now, the first mode is ask for approval. The next one

1:38is approve [music] for me, and the last one is full access. I normally like to put it on approve for me, as that sits in the sweet zone of the agent asking for requests. So, approve for me works just fine. And then down here, [music] you can choose the model and the advanced level you'd want. Typically, a medium level is just fine for this workflow. Once you're done with that, we now need to prompt this agent in order for it to do what we want it to do. Now, for that, I've created a document for you that has all the prompts we're going to use. Just comment the word prompt and

2:06you should have access to it in the comment section. What you've got to do now, once you have access to it, click on tab two, highlight everything on tab two, and copy it. Just copy it. And once you do that, head back to tab one and you'll see something called the engine prompt. Click on that. It takes you to a Google Drive. Just download that engine prompt from the document so it saves to your computer. And once you're done with that, head back to your agent, paste in the prompt you just copied, and just send it. Now, once you do that, the agent asks you if you'd like to install

2:35TopView MCP. What you've got to do now is type yes. Now, TopView MCP is pretty much just the backbone behind this whole operation. It gives us access to very advanced models like C Dance, Nano Banana, Cling, so many good advanced AI models. TopView MCP gives us access to all of them, all of them from the use of our agent. So, it gives our agent more tentacles to be able to use more things, helping it achieve a lot more when it comes to automated workflows. So, what you've got

3:03to do now is head over to your agent. For me, it's Codex. I'm going to type yes and then send it. Then the agent starts to check the tools it has and checks if the environment is set up correctly for TopView MCP. All right, once it authenticates, it tells you, "Done. TopView has been authenticated and you've been successfully connected." And what you've got to do now is just type yes, because it's asking you if you'd like to

3:27proceed. Just type yes and then send it. Now, once you do this, it asks you for the engine prompt, which is the PDF we downloaded from the drive. Just attach that PDF to the agent's prompt box, say something like, "Done," and then send it. So, the agent starts to read through the PDF. It starts to absorb the style of the PDF. And then, what it does next is give you a couple of niches you could pick from. Now, all these are just the niches that can be used for this Vox documentary [music] workflow. You've got things like history, money and power, crime and documentary, technology, sports, and so on. For this one, I'm

4:00going to pick the default, which is number one, crime and documentary. I'm going to just pick that and then send it. All right. Now, the next thing the agent does is give you a couple of video ideas in this documentary niche you've chosen. And all you've got to do is read through them [music] and then pick an idea you like. Now, for

4:16this, I think I like idea number three. So, I'm going to just pick idea number three and then send it. And when I do this, what happens next is the agent starts to ask [music] me how long I'd like the video to be. Now, what you could do, ideally, is pick a duration here. I want the video to be about just 1 minute long. I'm going to just send 1 minute. And when I do this, the agent now proceeds to create a really, really good script with this

4:40topic I've chosen. Now, the script looks pretty good. It follows the pattern of documentaries with a timestamp as the first sentence. And then it introduces the main character and just proceeds with the whole documentary flow. This looks pretty good. If you like how the script sounds, all you've got to do now is if you want your agent to generate the voice-over for you, just type voice. So, it uses ElevenLabs to generate the voice-over straight up for you. Or, if you wanted to just proceed to the next

5:08steps, you can just click on proceed. For this, I'm going to just say voice. So, the agent starts to create the voice-over for me. And when I type voice, it starts to check if I have ElevenLabs keys saved on my computer. Now, as you can see, Codex did a wonderful job in generating the voice-over for us. It sounds really high quality. I mean, this is really, really impressive. It downloaded the voice-over and saved it in the folder this whole session is saved in. So, you can also still save as a copy if you'd like to save this voice-over in a random folder,

5:41and it'll still save to your device. Now, what you've got to do next is just type proceed. And when you type proceed, what happens next is the agent now starts to break down your script into scene-by-scene chunks. It starts breaking the narrative [music] into beats and segments. And as you can see, it has broken down our 1-minute

5:59script into about 34 different beats. Now, according to it, it's breaking down the beats into 1 second. So, for every second, we've got a beat. And this looks pretty good. This looks pretty nice. You can also ask it to change the beats to be maybe for every 2 seconds. Like, for example, I'm going to say I want my beats to be for every 2 seconds. So, it rewrites the beats. If you like the beats it gives you in the first go, that's still fine. I've gone through the first go, and while it looks okay, I'd like mine to be a little bit restructured. Right now, we have about

6:3130 31 beats, and it looks like [music] the sentences are cut off at specific stamps in a sentence. And this kind of looks a bit messy. So, what I want to do now is tell it the exact [music] amount of beats I want. I'm going to say I want only 15 beats for the entire story. So, make a better segmentation beat division. Now, you can actually just tell it the amount of beats you want for

6:51your story. [music] Or, if you want to, you can also just read through the script and then calculate how many beats you feel exist in the whole script. For example, from here, December 2nd till his son could be an entire beat. And then from the Colombian police surveillance team till the street could be another beat. And then you could just keep dividing things like that. And then you [music]

7:13tell it where and what beats are valid. And it takes in the beat division you gave it and then gives you a new amount of output. And right now, it's [music] doing that for me. Yeah, this is a wonderful job. This looks pretty pretty good. I mean, it was able to take in the right chunks from our script and

7:28actually make really really good beats. So, I think the best route is to always tell it the amount of beats you want. So, it's able to give you a nice output. Now, this looks pretty good, so I'm going to just type proceed. And then what happens next is the agent starts to use the top view MCP we installed as a backbone for this

7:48pipeline. It starts to create the images, which are the start frames for the animation we're currently about to work with. All right, right now, as you can see, our agent has generated the images for us and it has given us the option to preview them and they look pretty good. They look pretty pretty good. Everything follows the style we're currently going for and this is really really impressive. However, it was able to generate 10 images in this go. And if

8:11you can remember, I asked for 15 beats. So, there's five images remaining. To continue, just type continue, so it continues with the next batch of images. And then we proceed with the next steps. But all you've got to do is grant it permissions by either saying proceed, continue, or very simple, minimal affirmative type permissions. All right. Now, after a couple of

8:32seconds, the output has been delivered. We now have the second batch of images. Full disclosure, I did have to try a couple of times because [music] I ran into issues with Codex. It was very frustrating. But yeah, we now have all 15 images we need for the animation and they all look pretty good. I like the quality and consistency. Now, what I'm going to do is just type next in my agent and then send it, so it proceeds with the video animation phase of this pipeline. All right, now after a couple of seconds, we now have the first batch of videos and they look pretty cool. I mean, the quality of this is

9:07really really impressive. It follows the animation style we're currently working with. It looks pretty good. All 10 video clips have been generated and yeah, it looks really high quality. Now, what I've got to do is just ask it to proceed with the next batch of clips. I'm not going to type [music] next, I'm just going to type continue, rather, so it

9:24continues with the next batch of clips. All right, now after a couple of seconds, we now have the next batch of videos. And yeah, everything pretty much looks pretty good. We have every single thing we need for the video. Now, one thing you could do is actually ask the agent to sync up the clips [music] and edit them with the voice over it created. But, to be honest, that would really kind of suck because sometimes the agent doesn't know when to fuse what

9:48clip to what chunk in the voice over. So, for that step, you'd have to use your own creative input. So, what you've got to do now is take the clips you've just gotten from the agent, the voice over and the video clips, take them into an editing software, and then just chop it together and edit it yourself. That's just the best route to get the best possible quality right now. Now, for the editing, I'm going to use an editing software called CapCut. It's a fairly simple workflow. Just drag the voice over you got from your agent down to the timeline, increase the volume, and then stretch out the timeline so it's wide

10:20enough to display your entire track. And then what you're going to do is get a background track. You can either get this from the CapCut library, or you can get it from YouTube Studio. It's best to get it from YouTube Studio so you don't run into any issues with copyright strikes. Just drag the background track to the timeline, cut the edge, add a fade out and a fade in animation, and then reduce the [music] volume. And once you do this, the next thing you've got to do is start dragging your clips to the timeline. Luckily, the agent has already labeled each clip in the exact order. So, just drag them onto the

10:48timeline in the exact order you [music] got them. Now, you might need to speed up these clips to play it like two times because when you're done dragging all the clips to the timeline, there'll still be room for a lot more. So, after dragging everything to the timeline, just highlight the entire clips, go to speed, and increase it to about maybe two times so it plays a lot faster and it's just in sync with the voice over narration. Once you've aligned everything, you [music] can now play

11:11with transitions. You can just drag in some transitions to the timeline in between two clips to just make sure the change of scenery is smooth. Now, you don't need to play with too much transitions since this kind of animation and collage effect is already pretty cool. And what you've got to do now is just preview everything, and if you like how

11:29it looks, just export the video. And that's pretty much it.