45 Seconds of You. AI Does the Rest. Short Form Content with n8n with Adam Goodyer

🎁  Download n8n Template 1 Here -https://go.n8n.io/ai-short-form-content
🎁  Download n8n Template 2 Here - https://go.n8n.io/human-short-form-co...

What if you could turn one long form video into 3 fully edited shorts — with scripts, B-roll, captions, transitions, and a lead magnet — all automated?


Timestamp
0:00 - Introduction & what we're building
0:04 - 50,000 LinkedIn impressions from AI-generated content
0:14 - Overview: Long form to short form with minimal human input
2:42 - Phase 1: The fully automated approach (AI avatar)
2:56 - Building AI clones with HeyGen
3:01 - Using Appify to scrape YouTube videos & transcripts
3:37 - The problem with AI avatars & why Adam iterated
5:13 - Live demo of the AI avatar short (example output)
5:45 - B-roll extraction from long form video
15:17 - Building a template library for animations & transitions
18:22 - The AI Video Director node explained
21:20 - Automated lead magnet generation
22:37 - Phase 2: The Telegram approach (human + AI hybrid)
26:34 - The full content pipeline vision & what's next
28:12 - How to set up the workflows (API keys walkthrough)
30:52 - Where to find Adam & his agency

Get 30% Off n8n Cloud Starter or Pro Plans!
Want to get started with n8n? Visit n8n.io/pricing and use code 2025-N8N-PODCAST-729C416E at checkout for 30% off your first month or year.

Transcript

SPEAKER_00 0:00

I'm a bit of a lazy guy when it comes to short form content. I much prefer long form. I think I've had close to 50,000 impressions on the LinkedIn content that we're generating. Largely written by AI. It's good, but there is that bit of like, I am an opot, I am talking like this. Today we're gonna be talking about creating short form content on N8N from long form content with very minimal human input.

SPEAKER_01 0:23

Adam built a system that turns one YouTube video into three fully edited shorts. Scripts, AIB roles, transitions, captions, and even lead magnets, all compiled and pushed into Google Drive automatically. But when the AI Avatar wasn't cutting it, he iterated. He built a second version that pings you on Telegram with a script. You film yourself in 45 seconds, send it back, and the system handles the rest. Humans, where it matters, AI, everywhere else. Full workflow template in the description. Let's get into it. Adam, so what are we learning today, brother?

SPEAKER_00 0:58

G'day, Dylan. So today we're gonna be talking about creating short form content on N8N from long form content with very minimal human input. So we're gonna automate a lot of those steps.

SPEAKER_01 1:10

Fantastic. Let's get into it. Open up your screen and we'll dive into it. I know as a content creator myself, you know, the hardest thing is not only making the first long form piece of content, but then distributing it across all the different channels. So whenever you can do that and make that life easier, um, I'm all about it.

SPEAKER_00 1:26

Yeah, 100%. And, you know, especially for busy black people, a lot of people in the AI space, they're running agencies, they're doing more than just uh, they're doing more than just creating content. And so a lot of people I know are just they run out of time. They make a lot of excuses for being able to make that content across all these different social channels. So if we can use AI to, you know, we still have the human aspect of you actually teaching something or doing some sort of value, but we can use AI to come in and help with a lot of the, you know, the other things, the content research, the editing, that kind of stuff. It just gives time back to, you know, busy entrepreneurs, founders, whatever the case is, to be able to still focus on, you know, building a personal brand and creating content at the same time.

SPEAKER_01 2:07

Fantastic. And as you walk through this, let's walk through and also walk through a bit of your logic. I know you had a couple of different iterations on this one, and you kind of landed on a new version. So I'd love to talk about kind of how you iterated through this and as we go through this process.

SPEAKER_00 2:20

Yeah. So in this video, uh, so initially the way I thought of building content was, and I'm to be honest, I'm a bit of a lazy guy when it comes to short form content. I much prefer long form because I can do something like this where I can spend 20 minutes talking through exactly what's happening at each step. Whereas I think short form content is a lot more tiered towards hooks and it's quite a different content strategy, one that I'm just not great at. So my initial thought was well, let me build a completely automated solution to go from the long form videos that I'm creating on AI topics directly into shorts using HeyGen for using an AI clone of myself. So I made multiple AI clones of myself. Uh, then I used Gemini to analyze my long video. I extracted the transcript. I used Apify to actually scrape and download that YouTube video. Uh, and I had this whole process of going through and then using Creator Mate to uh to compile and render that video. And I think I think that might and then I was I was using replicate for AI B roll as well. So I was I was using a C dance model there. So I had this like big flow and and the output was good. I can show you what the output looks like in a moment. The output was good, but then the issue I had with that was I I don't want to make it look like I'm putting zero like if if you if you can tell that it's an AI avatar, it risks damaging your personal brand, in my opinion. So I the iteration was well, what if I try and automate everything else except the actual recording and use tele use a telegram note or something to just flick me a message with a transcript, I use a teleprompter app and film it there. Uh and so that that aspect is human and everything else is done by AI. AI is editing, it's compiling the whole video. Um, to still maintain that human element of it was was where the iteration has. So I'm still in two minds about like the best way to do something like this. But yeah, long story short, I got it all set up with Hey Gen uh and it was looking good. I mean, you can tell me whether or not you think it looks realistic or not. I think I think it was very close, but maybe it lacked some of the vibrant human movement that you know it was a maybe a little bit monotone, we'll say.

SPEAKER_01 4:24

Sure. Let's hit play. So are you gonna are you gonna show the video of the uh original Hey Gen version?

SPEAKER_00 4:31

Yeah, so this this is all compiled. So this uh the avatar in these videos, I can show you two videos. Sure. Everything in here is completely automated from start to finish. So the inputs are a YouTube URL, uh, and then the output was three different shorts having three different angles. So I had a creative angle, like with short form content. I think it's very much about an emotion. Like you'll see a lot of the content is like new tool, it does all these cool things, like that's a style. Or if you're not using this tool, you will be fired in six months, or like there's there's all yeah, so there's fear, curiosity, whatever the case is. So let me know if you can see this tab. I see it. Go ahead and play it. Cool. Let me know if you can hear it. Most people think Claud is just an chat GPT, but they just released something that changes everything. It's a new feature called co-work. It gives the AI direct access to your local files. It's not just reading them, it writes, moves, and edits. This is a huge step for AI agents. Think about all your scattered meeting notes or research for a new project. Instead of copy-pasting everything into a chat window, you just point the AI to your notes folder, tell it to create a summary of everything inside. It reads all the different files. So you can kind of then it writes a clean. See the point there. So it's taking, it's extracting B roll from an actual long form video that's relevant to what's being said. Um, this is obviously my AI avatar down the bottom here that's reading the script. And so it's it's come, and then it's got a number of different scenes as well. So it's got like we call this the 50-50 split.

SPEAKER_01 6:00

It's good, but there is that bit of like, I am an RBO, I am talking like this. So yeah. Yeah. Okay. All right. That's I mean, it's decent. Is it's not quite there yet. Um, maybe a couple evolutions, it will be, but okay. Um, we'll do a couple just do a couple seconds of this one, and then we'll we'll we'll jump into everything else.

SPEAKER_00 6:18

Your AI agent is probably burning cash. Every time it runs, you pay for tokens. Most people stuff the context window with data. They use vector search to find relevant chunks, but this brings in a lot of noise. The AI has to search. So I think it I think the voice is pretty good. So 11 Labs was done for the voice. I think the avatar, and I think this might just be a limitation with Hey Gen, to be honest, in terms of like realistic movement. I think there's a little bit of, you know, maybe two or three months left before those avatars are truly realistic, in my opinion. But in the in the short term, I think that means that, you know, human, and that that was kind of why we made this the switches. I did a lot of playing around with Hey Gen, trying to tweak it, trying to get it to look realistic and have a bit more, I guess, vibrance and human emotion in it. I just don't think it's there yet. It's from a tech perspective, but it's only a matter of time. Sure.

SPEAKER_01 7:12

All right. So that's fantastic. That was phase one. So let's let's take a look at phase two of what you what you what you built.

SPEAKER_00 7:19

Cool. So um we have now up quickly, there's there's one workflow that runs prior to this one, which is this workflow here. Now, there's two there's two ways to go about this. One is you could just upload a file if you had it locally, but I wanted to use it, I I wanted to just use get it straight off of YouTube because for me, long form is like the starting point of all my content. I'll film a 30, 40 minute video, whatever the case is, on some sort of tech topic. And then the what I would do is I'd come here and I'd drop the YouTube URL. Uh, we'd use an Appify scraper to download that YouTube video and the transcript. So we'd have those two things. Uh, and then we'd use a Gemini node. I only discovered this probably a month to two months ago. Um, it's crazy how powerful Gemini 3 Pro is in terms of analyzing videos and the kind of data you can get out of that. Insane.

SPEAKER_01 8:13

I could also see, just as a random thought, a topic is the analyzing videos. I also wondered if you could also use it for like sales meetings or other situations where you're analyzing how your mannerisms are, your communication, your cadence, and then coming back with like analysis beyond just the transcript.

SPEAKER_00 8:28

Yeah. That's cool. Well, because one thing I've also built is for clipping videos. So even through this meeting right now, what what you can do is if you have a transcript, you have a visual analysis, and you have a uh like waveform, audio, audio decibels to understand, you know, the volume at certain times, you can almost clip up entire videos in in less than 30 seconds by running a script because it can combine those three things and go, well, you were silent at these points. Um, you know, you were saying, like at this point, you were talking about some mumble jumble that wasn't relevant. You can use the combination of those three, and maybe at this point you were like drinking from your water bottle or you weren't even sitting down from the visual analysis. So the the combination of those three data points is all you really need to go ahead and be able to clip up entire videos in one click. So even that, like video editors spend hours just watching through like hours of content and trying to clip it at the perfect times. That can almost that can be done program programmatically now.

SPEAKER_01 9:24

Fantastic. Okay, cool. So you're analyzing the video, it's coming through, then it hits it's parsing the the key actions. What's happening inside the key actions?

SPEAKER_00 9:32

Yeah, so these code nodes are essentially just formatting, extracting the data and and basically just parsing the JSON that gets responded here to get that gets returned here. Uh, and then we have this AI section and also I can show you the system message for this one. So essentially what this is doing uh is it's extracting key data out of here. So what we're trying to do with this node is because the end goal was we could use this node to not only build shorts, but then later we can build more sub workflows for creating LinkedIn content, creating X content, um, whatever the key, you're creating content across all these different channels. If we can accurately analyze a video and be able to pull out all of the metadata, the topics talked about, the key, key points, all that kind of stuff. We could feed that not only into shorts, but into every other social content form as well.

SPEAKER_01 10:17

What I like about that is the fact that you can some people they'll take a short and they'll just use like repurpose that IO and they'll send it to a bunch of different channels. But what works on TikTok doesn't quite work on LinkedIn. Yeah, it's a very different messaging, communication, cadence style. So if you had the ability to switch out the way it landed on those different channels, you would you're I'm sure your um uh reach would go higher.

SPEAKER_00 10:43

100%. And I've had that because I've been doing a lot of link, even in the last week, I think I've had close to 50,000 impressions on the LinkedIn content that we're generating, uh, largely written by and the content strategy there is completely different to what you would use on something like Instagram. Pole polar opposite almost. So yeah, 100% there's there's there's that in there. So once we have that information, we could pass that into a workflow.

SPEAKER_01 11:06

So you're pulling up the workflow right now, and I could see the benefits of of analyzing a long form video, slicing it up, not only in shorts. And also, I'm sure this was mentioned the very beginning of the podcast, but these workflows are gonna be made available so that you can be able to download them, get started, and iterate with them. So anybody else that wants to be able to do this, that's listening to this, will be able to do the same thing without the mass amount of uh uh time put in for you to create it.

SPEAKER_00 11:28

Exactly. Now, this is the original workflow. I just found it now. I've I've got a number of different workflows in here, and I'm like, which ones which? That's a that's a key to be naming your workflows better than I name them because mine have got a lot of random names. Um, so what would happen is that short trigger uh gets actioned. Again, a lot of these are just setting variables. Um, but then we'd have this short form content creator. So, like I mentioned before, you'll see here now a couple things. So, one, sorry, what happens here is we I define the user prompt, the system prompts, and the examples. Now, there's some very prominent names in the AI space doing better than I am, creating really dope content. And I didn't want to reinvent the wheel on, you know, especially for short form where hooks and that kind of content style is so important. I go, I thought, let me go and look at you know what the top performing content creators and my niche are doing. Let me grab the captions, the transcripts. Again, I used Apify to scope the transcripts and the the most engaging posts. Let me use those to feed the agent that's generating the scripts for mine to mimic that kind of content style. So that's what you're seeing here. I've got you know, Nick's obviously doing a lot of things. I think Nate's, yeah, Nate's in here as well. Uh, a couple of the top top people in here, the transcripts, all that information.

SPEAKER_01 12:42

Shout out to Nick and Nate. What's up, guys? Shout out to Nick and Nate.

SPEAKER_00 12:45

Shout out to Nick and Nate. Um, and so what this is doing is uh essentially it's got a lot of my brand information, but what it's doing is it's building the scripts and content ideas, like we said before, for three different angles. So we had the fear example that now it's a very big prompt. Um, you have to download this to kind of go through it all in detail, but it's essentially building the scripts from I think a curiosity angle, uh I can't even remember, like a fear angle, three different angles. Uh, and it's then sending that through here, and we're looping through each of the three concepts. One, we're getting the AI B roll. Now, this was the C Dance One Lite model. The reason for that is because I just wanted basic AI B roll, and this is cheap. I think it was uh it's a couple of cents a video, maybe even less than that. So we're using uh through replicate, we're using for two, three, three two-second reels, AI B roll. Um, so we're generating with our first concept, we go through, we generate the B roll, um, we build the Creator Made Effects library. Now, one thing with AI is that if for editing a short form video or editing videos in general, pause for one hot second.

SPEAKER_01 13:51

I do want to say, so for C Dance is created by TikTok, and they had Seed Dance 1.0, which came out, which was pretty good, but they are they came out with C Dance 2.0 that has also come out, which is um brand new to the market, at least by the recording of this podcast, and it is is orders of magnitude better. They had some really good things that they did. There were some copyright issues where they could do Disney movies in Star Wars, and fleets of uh litigation came their way. So they might be pulling out the ability to do that. But the seed dance is one of the latest and greatest, at least the 2.0 version, for creating this type of AI-generated video.

SPEAKER_00 14:27

Yeah, cool. I so yeah, and this I probably built this like a month ago around then. So yeah, it could well be worth. And there was a million models as well. I remember looking on replicate, and I was like, there are so many random models everywhere, they all had different pricing. I was like a bit overwhelmed, to be honest, on exactly what the best one would be. But um, yeah, that could that could definitely be worth a look. So the next thing with um create editing videos, this one's actually really important because I'm not sure if you a lot of people are talking about remotion as well, and like there's a lot of AI, AI to edit videos is quite a popular topic right now. And what a lot of people think is that they can just pass it a video uh and then be able to edit video, like an AI can edit a 60-minute video really easily. What you want to do is instead of letting the AI just dis decide the animations, the transition styles, the captions that are wraps on there, you want to build a template library of animations, transitions, b-roll moments, etc., that look that you're happy with that match your brand. And then instead of making the AI do all the heavy lifting and not only analyzing the video and and identifying where and how to edit it, if you make the AI also think about, you know, what animations to use, it's more heavy lifting. And we want to try and move as much of the heavy lifting into like hard-coded templates and then just have the AI instead of having to generate the animations and then apply them, it just needs to pick out of an already pre-built library. These all work, we're happy with how these work. It just needs to like shift them and move them into the positions that work. So it's less heavy lifting from the AI, but there's less that can go wrong, and you get more reliable results if you do it this way.

SPEAKER_01 16:05

And to let people know, uh, Creator Mate, which is what he's talking about, is essentially a programmatically uh video editor platform that allows you to feed in whatever you'd like. And programmatically, you can actually make these videos. And so what you're saying is you want to use this platform, Creator Mate, that you're using to and teach it and offload it on like this is how you can edit and move in B-rolls and transitions and all that types of stuff, so that uh the AI isn't really trying to one-shot a perfectly generated UGC video by just feeding it to the video uh seed dance or any other type of AI video generated platform. You're actually you're actually using an actual program platform that can do that along with the B-roll and or any of the other content coming from it to be fed into that system. Is that correct?

SPEAKER_00 16:53

Yeah, exactly. Exactly. Yeah. And so all of like the scenes and everything, we've already got pre-built in here. And if we wanted to change it later down the line, we wanted to change our caption style instead of having to like go through all these nodes and change it in, we just change it in the effects library. This is like our templates library, and we use this node when we go to actually compile the video later down the line. Nice.

SPEAKER_01 17:14

And I could see adding like a set node or something or some way to be able to easily get inside of there to be able to change this out to make it easy of like, hey, match my brand, and then just and then you just update that set node, maybe hooked up to like a uh N8N form or something else that you could just populate on the front end. So if this was made available for a client, you say, hey, pop in your new information, that will automatically bring it into the the set node function that will then update this um uh this creator mate node that has all the relative information if we're gonna like build this out for a client.

SPEAKER_00 17:46

100%, 100%. And you could have different yeah, screen styles, and you could you could definitely build that into a set node.

SPEAKER_01 17:52

Beautiful.

SPEAKER_00 17:54

And so once we have so first we've got we've got our concept idea, we've got the AI generated B roll, we've got three short two-second clips, and we've built our template of creator mate scenes. We then go ahead and build the actual AI avatar video using the script that was generated for this concept. So typically the scripts are anywhere from 30 to 60 seconds. So we basically just loop through and poll that video until it's done. And then once we have our avatar video, we send it to the AI video director. So essentially, this is this is what's actually going ahead and building the script. It has the HeyGen video, which is the actual AI avatar talking. So it knows exactly how long the short should be. It's got an idea of the long form video and where the B roll could be extracted, because we did that in the previous workflow that I showed you, which is which we actually do across every social media content strategy. It's got the AI B roll uh and it understands the creator mate template library. So what this is going to do is it's going to output this like long JSON structure, which long JSON storyboard structure, which will be broken down into each scene. So like scene one could be four seconds, and that's the hook, and that's you know, that's full screen avatar. Me saying, hey, look, look at this new tool or look at this new AI model that does this. And then it will scene two could be B-roll, C three could be like a 50-50 scene split. So it's it's using this structured output pass-up, and we can actually take a look at what this looks like to uh output all that data in the exact JSON structure that we know is going to work with Creator Mate so that we can then go in and basically there's a little code node to like adjust the body to make it formatted correctly for Creator Mate, and we can send it off to Creator Mate to actually render that video. Again, it polls that uh and then it upload updates it straight into uh Google Drive so you can then access that video to go and schedule it wherever you want to use it. Um, and you can also use a social media copywriter to build uh descriptions if you wanted to kind of uh so we use it to build the caption for Instagram, the video title, etc. etc. But you can have those little tweaks in there to append that row into a sh to a Google Sheet, uh, which is a central spot that you can go and just grab all your videos, all the content, stuff like that. And it's a little loop through that uh and create three videos. I mean, you could easily change this to make 10 videos if that you that's what you wanted. You can loop through an infinite number of videos, but I make three. Uh and then at the very end, this is actually a cool thing as well, was I generated a little lead lead magnet. So based on the idea, obviously you'll see on Instagram a very common comment style is like comment X for Y. It'll be like, Comment, there's this new tool or this new model. Um if you want to try it out, comment this and get it.

SPEAKER_01 20:40

Engagement farming.

SPEAKER_00 20:41

Yeah, exactly. Exactly.

SPEAKER_01 20:42

Engagement farming for anybody that doesn't know, uh if you're not part of the world, is uh before people would say, hey, DM me and I'll respond. But if I if you ask the person to comment down below, then the algorithm will see all of a sudden hundreds or thousands of comments about the word optimize or telegram or whatnot. And then that that boosts the algorithm so that you get more of a reach. But if you make a if you if you want to make a good lead magnet part of the video, then this is gonna use AI to generate that lead magnet for you. Yep, exactly. Fantastic.

SPEAKER_00 21:13

Exactly. And it's actually it's got like it's there's a little, so you can see this document to HTML. So this generates the copy, and then this has a it's got like a little PDF format. It's got all my, you know, logos for my YouTube channel, my social media. So like they go to the lead magnet and they get they can see that kind of nicely formatted document there.

SPEAKER_01 21:30

I could also see if you if you iterated on this type of thing and you hooked it up to say a mini chat environment for Instagram. Instagram is is a very favorable automated DM response system. And so you could programmatically hook that up to be able to have mini chats, look at the word that was used, the the the whatever's going on inside of the video, and then make sure you send the right lead magnet over automatically.

SPEAKER_00 21:52

Exactly. Exactly. And and this is appending in the same Google Sheet as well. So it's all it's all in that one. So you Exactly. Like we actually set that mini chat automation up as well. So that you can um all the information's in one row in Google Sheets. And that works really well with setting up further automations to be able to extract the data when people comment certain things and stuff like that.

SPEAKER_01 22:13

So you said this is is this currently hooked up to Hey Jen? Because you're talking about another model.

SPEAKER_00 22:18

It was currently hooked up to Hey Gen. So I'll sh quickly show you through. So we I'll show you quickly through this one. Oh sorry, that's not one. It is that one. So the Telegram approach was uh yeah. So I've called this one short form creation human. Uh so this is the telegram approach. So the telegram approach stripped the HeyGen stuff out of here. And instead, because what I what I wanted to do is for me, it's mainly about time savings. I'm happy to I love making long-form content. I'll be honest, I don't enjoy making short form content that much because I just don't think you can offer that much value in 45 seconds of someone's attention span. And it's more just about like hooks and and trying to engage people. Uh so what I was thinking was if you if if we can build a script which literally just pings me on Telegram and goes, hey, here's the script, record the script. That's the only thing that I need to do. I'm happy to do that and make it a little bit more human. Uh and it's it's just a bit more genuine because I'm I'm actually recording myself. So this is the same thing that we had there, but stripping out a lot of the AI features and actually just using Telegram to uh to pass the they are to be able to film the video and have an easy to use chat interface where I can go straight from my phone back to a computer. So you can see here we're using Telegram nodes to ask the user for input, generate the concepts, ask the user to pick a concept. So it'll give me a couple of options about what I want to talk about, ask the user to film the video, then it converts that video and basically sends it to Gemini for analysis. Again, and then it follows a similar process to what we had before, where it's going to compile that video, uh build a storyboard, which I think is an essential step, and then convert that storyboard into like creator mate syntax and sends it off to creator mate for editing.

SPEAKER_01 24:07

Yeah, I could see. Well, question about the telegram. Does the telegram message you and say, here's your script, and then you read it back? Is that what's happening?

SPEAKER_00 24:15

So, yeah, so you can copy the script into a teleprompter app. So that's what it says. It will first it'll say, hey, here's three concept ideas from three different angles. Which one do you like? And you can go one, two, or three. Uh and then it will go, okay, here's the script. Um, and so you'll grab the script, you'll copy and paste that into a teleprompter app. If you want to use a teleprompter, if you want to just memorize it and do it off the top of your head, you can do that as well. Record that video uh and send it back. So yeah, it's like super quick and easy to go to do it on your phone, which is obviously optimized for familiar short-form content.

SPEAKER_01 24:45

What I love about this is the fact that you focus on the human element of you presenting the information. And because there's gonna be a script that you get back that you may not read exactly, but you understand the general gist of. And there might be some words or vocabulary that you use that would make you sound more human. So I love putting the human element back into it and just focusing on the you said having the human in the center and building systems all around the human where you just come in and you just you just put in here's the human face, my voice, my tonation. Uh, tonation? Don't know if that's a real word. But but my tone to be able to inclination to be able to have the ability to speak this out and in a way that would actually land and resonate so people would feel it, because you do get that uh un, what do you call it, uncanny valley effect that will happen when it's when it's just the AI reading you feel something slightly off. So yeah, fant fantastic. Where do you where do you plan to take it after this? Like, what's the thoughts on the evolution of this?

SPEAKER_00 25:45

The the thoughts are to to build again, to build an entire content pipeline. Like the idea is that I can build a entire content pipeline that helps me ideate video, ideate concepts. So I guess first step would be research, um, you know, ideation and research through to building, you know, video ideas both for long form and short form, and then being able to repurpose that. So essentially reducing the time my content creation workflow time, because it's it's a serial time investor for anyone out there that's running, you know, that's doing YouTube channels, Instagram, TikTok, LinkedIn, X, building uh marketing campaigns, like that, it's a full-time job. And so being able to try and do that whilst also trying to run an agency uh and whatever other ventures you're involved in, it's it's borderline not possible. So the idea is to build a system where we can maintain the human aspect and I can talk about everything that we're doing at the agency and and offer all that value, but then have a system that's going to mean that I can still do those things and not have to, you know, invest in being a content creator full time. I can use AI to help out with all the manual stuff, the video editing, the ideation, the competitor research, the thumbnail generation, you know, all those things that are going on behind the scenes. Uh can we can we you know maintain the human aspect of it, but still but use AI to help out with all of the repetitive processes essentially and just reduce the time it takes down by 80% is the is the goal.

SPEAKER_01 27:11

Makes total sense. I love it. Uh so this workflow will be available for people to download inside of here. Uh you can go click the link. Um we'll have that video embedded for you. And then I get a question for the creator made library of those embedded libraries. That will that all be available as well?

SPEAKER_00 27:28

Yes. So that's all in this Creator Mate Effects library node here. So you can uh these you'll see is a fair bit of like JSON in here, but it's actually quite it's it's all um, I've even got some like audio tracks and stuff in here. Uh I'm actually hosting them myself, but everyone can use them. Uh but yeah, you'll see there's like these are just like you know, transitions, like flashes and all that kind of stuff. There's it's just a big library of all these different effects that you can reuse essentially.

SPEAKER_01 27:55

Makes sense. So if someone was to download this, uh what would be the the steps to setting this up? What would be like the ABC steps?

SPEAKER_00 28:02

Uh pretty much. So there's two, there's two, it depends how you want to do this, um, but there's two workflows. So the first one's gonna be this one here, which is going from YouTube. Now, the main thing is just gonna be swapping your API keys over because we are using a couple of third-party services. We're using Apify. Um, you're gonna have to throw your Gemini API key in. Of course, you could change these LLMs, like Gemini is the only one that analyzes video to my, but you could change this over to Gemini if you wanted to, and I think that'd be fine. Uh, but otherwise, you'll have you have to throw your API keys in here. That's the only thing you'll need to do for this node is to just set up your API keys. For this one over here.

SPEAKER_01 28:41

Got it. So so Appify, Gemini, OpenAI.

SPEAKER_00 28:46

Yep. Okay. And then for here, you will need you'll need the replicate. Uh so you'll need Hey Gen if you want to go down the HeyGen approach. Um, if you want to take the Telegram approach, you'll need Telegram, uh, Google Drive. That'll uh Google Drive if you want to create the lead magnet as well. Um, that'll be the same API key for Sheets and G drive, as long as it has the correct permissions. Uh, and then you'll need replicate. Um, this one here, if you want to use C Dance, you you can definitely upgrade this to the C Dance 2 model that you're talking about as well.

SPEAKER_01 29:18

Replicate is a platform that hosts a series of different uh AI models that allows you to hot swap between them, kind of like an open router, but this does video generation, correct?

SPEAKER_00 29:30

Exactly. Exactly. And so maybe it would be cheaper if I'm not sure if you can go directly to the C dance, like the LLM provider to get a key. I'm not sure if that's how it works.

SPEAKER_01 29:40

I'm not too sure right now. And but but okay, cool. So replicate and then yeah, agent and create a mate. Create a mate.

unknown 29:48

Okay.

SPEAKER_01 29:49

Yep.

SPEAKER_00 29:49

Absolutely. So literally just swap the API keys. I don't think you're gonna need to do NVIDIA anything else. And then you obviously have to create your own Google Sheet as well. So it won't be linked to my Google sheet if you if you want to have that in there as well. Um just hard code your Google Sheet in there, man. Just fine. Hard code, yeah. Everyone can use my Google Sheet.

SPEAKER_01 30:08

Amazing. You could also use data tables if you're feeling so curious. If you if you don't want to create uh that account, uh the internal data tables for in it and works. Um but okay, fantastic. Well, um, Adam, it's been awesome having you in the show. This is uh really valuable. I think is really needed in the time of like, you know, getting outside of getting stuck from the inside the world of watching content and then being more of a producer of content with your own flair. And I can imagine people could evolve this to even more where they could add in their own tone, their own voice, their own style, uh, and and be able to upgrade this system so that it's coming from them as a person, you know, based upon their interests and what they're into. So 100%. Fantastic. Uh so this is amazing. Thank you for having been on the show. If people want to find you and find out more about what you do, how do they do that?

SPEAKER_00 30:56

Yeah, so I'm I run a YouTube channel. Um, you can come Adam Goodge's YouTube channel. I post a lot on LinkedIn as well. If you're interested in, so we we pretty much do everything AI, but if you're interested in like learning how, you know, how to sell high-ticket AI so it's full stack software, um, you can take a look at our agency, which is APG Software. Um, you can actually review a live demo of like the kind of internal systems that we build. So we build internal systems for small to medium enterprises, basically help them get control of their data and adopt AI. So if if people are interested in seeing what that looks like as well, they're more than welcome to go and take take a look at that.

SPEAKER_01 31:30

Amazing. Adam, thank you so much for being on the show. It's been an honor and a pleasure, my friend. Much love, and I'll see you on the other side.

SPEAKER_00 31:37

Likewise, mate. Appreciate it. See you on the other side.