WEBVTT 00:00:00.000 --> 00:00:04.080 Imagine for a moment a world where those really 00:00:04.080 --> 00:00:07.320 captivating, sometimes oddly specific viral videos, 00:00:07.480 --> 00:00:10.380 you know, the ones are actually conceived and 00:00:10.380 --> 00:00:13.439 created not by a person, but by an A .I. like 00:00:13.439 --> 00:00:15.720 a tireless digital artist. Yeah, exactly. Like 00:00:15.720 --> 00:00:18.059 slicing crystal fruit. Yeah. With that weirdly 00:00:18.059 --> 00:00:21.239 satisfying precision. And, you know, it's not 00:00:21.239 --> 00:00:23.239 science fiction anymore. This is actually happening 00:00:23.239 --> 00:00:26.289 right now. Welcome back to the deep dive. Today, 00:00:26.390 --> 00:00:28.649 we're really going to unpack a fascinating blueprint, 00:00:28.929 --> 00:00:31.929 a fully automated system for AI video creation. 00:00:32.090 --> 00:00:34.210 Think of it maybe as your own digital factory 00:00:34.210 --> 00:00:38.289 for potentially viral content. Our source material 00:00:38.289 --> 00:00:40.750 is this really detailed guide, a genuine step 00:00:40.750 --> 00:00:42.950 -by -step for building what it calls an AI video 00:00:42.950 --> 00:00:45.250 factory. We're going to explore exactly how those 00:00:45.250 --> 00:00:48.189 kind of captivating ASMR glass cutting videos 00:00:48.189 --> 00:00:50.789 can be produced at scale. Yeah. And our mission 00:00:50.789 --> 00:00:52.329 today, it's really to show you how this whole 00:00:52.329 --> 00:00:54.770 system works from an AI basically dreaming up 00:00:54.770 --> 00:00:57.189 ideas all the way to videos being published across 00:00:57.189 --> 00:00:59.649 social media. And importantly, with minimal human 00:00:59.649 --> 00:01:01.789 touch. We'll walk you through this setup first, 00:01:01.929 --> 00:01:05.329 then how the AI generates unique concepts, how 00:01:05.329 --> 00:01:07.329 those concepts actually become finished videos. 00:01:07.409 --> 00:01:10.790 And then we'll dive into the economics and maybe 00:01:10.790 --> 00:01:13.530 the deeper philosophy behind doing automation 00:01:13.530 --> 00:01:16.469 smartly. The Internet is truly a strange place, 00:01:16.530 --> 00:01:19.480 isn't it? Millions of people just. captivated 00:01:19.480 --> 00:01:23.280 by videos of knives slowly slicing through glass 00:01:23.280 --> 00:01:26.019 fruit, like a crystal banana or a glass apple 00:01:26.019 --> 00:01:27.840 being cut. There's just something incredibly 00:01:27.840 --> 00:01:30.900 satisfying about it. And for content creators, 00:01:31.140 --> 00:01:33.400 this is a huge opportunity. I mean, the digital 00:01:33.400 --> 00:01:35.219 landscape is so crowded, right? Attention is 00:01:35.219 --> 00:01:38.079 the currency. But trying to produce visually 00:01:38.079 --> 00:01:41.099 complex stuff like this every single day, man, 00:01:41.159 --> 00:01:43.010 this is a recipe for burnout. Think about the 00:01:43.010 --> 00:01:44.950 manual work involved. You've got the lighting, 00:01:45.010 --> 00:01:47.430 the sound design, the really intricate editing. 00:01:47.569 --> 00:01:49.370 It's just an enormous amount of effort for one 00:01:49.370 --> 00:01:51.510 person. Right. And this is where clever automation 00:01:51.510 --> 00:01:53.969 really changes the game. It transforms that manual 00:01:53.969 --> 00:01:56.569 craft into a, well, a scalable system. Yeah. 00:01:56.590 --> 00:01:59.250 We're essentially hiring like a tiny digital 00:01:59.250 --> 00:02:02.450 video artist, one that works 24 -7, just dreaming 00:02:02.450 --> 00:02:06.109 up beautiful glass things and making these captivating 00:02:06.109 --> 00:02:08.389 slicing videos. So before we even start building, 00:02:08.530 --> 00:02:11.319 we need to prepare our... ingredients and lay 00:02:11.319 --> 00:02:13.860 out the tools this initial setup it's mostly 00:02:13.860 --> 00:02:16.400 a one -time thing but doing it thoughtfully makes 00:02:16.400 --> 00:02:18.879 the whole process way smoother down the line 00:02:18.879 --> 00:02:21.900 okay first up Your automation platform. This 00:02:21.900 --> 00:02:23.400 is going to be the factory floor, the central 00:02:23.400 --> 00:02:27.180 hub. Tools like N8n or Make .com are great. They're 00:02:27.180 --> 00:02:30.419 visual, so you connect applications like building 00:02:30.419 --> 00:02:32.680 blocks. Why visual, though? What's the advantage? 00:02:33.099 --> 00:02:35.400 Well, instead of staring at lines of code, you 00:02:35.400 --> 00:02:37.819 actually see how the data flows. It makes building, 00:02:38.020 --> 00:02:39.960 and maybe more importantly, fixing things later, 00:02:40.120 --> 00:02:42.680 much more intuitive. Okay, makes sense. Then, 00:02:42.740 --> 00:02:45.139 the memory bank. Our AI needs to remember what 00:02:45.139 --> 00:02:47.340 it's already made, right? To avoid just repeating 00:02:47.340 --> 00:02:50.699 itself. Exactly. And a simple, free cloud spreadsheet 00:02:50.699 --> 00:02:53.159 like Google Sheets is perfect for this. It's 00:02:53.159 --> 00:02:55.340 accessible everywhere. It connects easily to 00:02:55.340 --> 00:02:58.740 other services through its API. And, you know, 00:02:58.759 --> 00:03:00.659 a human can easily check it too. Just set up 00:03:00.659 --> 00:03:04.580 two columns, object and video URL. Simple. Got 00:03:04.580 --> 00:03:08.539 it. Next, the AI brains. We need access to a 00:03:08.539 --> 00:03:12.240 powerful, large language model, an LLM. That's 00:03:12.240 --> 00:03:14.280 the AI that can understand and generate text 00:03:14.280 --> 00:03:17.240 that sounds human. Right. Think OpenAI, Anthropic, 00:03:17.240 --> 00:03:19.699 Google AI, services like that. You'll need an 00:03:19.699 --> 00:03:21.860 API key from one of them. Yeah. Think of it like 00:03:21.860 --> 00:03:24.020 a secret key you give to a trusted delivery person, 00:03:24.159 --> 00:03:26.620 lets your platform talk to the AI. Right. The 00:03:26.620 --> 00:03:29.300 source we looked at uses OpenRouter, which is 00:03:29.300 --> 00:03:31.400 neat because it unifies access to lots of different 00:03:31.400 --> 00:03:33.680 models. And then the actual video production 00:03:33.680 --> 00:03:36.520 service. This is the specialized AI that creates 00:03:36.520 --> 00:03:39.199 the video. Mm -hmm. And using a middleman service, 00:03:39.539 --> 00:03:42.080 something like Foul .ai, is often easier than 00:03:42.080 --> 00:03:44.080 connecting directly to the big models like, say, 00:03:44.180 --> 00:03:47.020 Google VO3. Why is that? Why the middleman? Well, 00:03:47.199 --> 00:03:48.879 they handle all the tricky server stuff, the 00:03:48.879 --> 00:03:50.939 management, and they give you one consistent 00:03:50.939 --> 00:03:53.080 way to order a video. Makes life simpler. You 00:03:53.080 --> 00:03:56.180 just grab their API key, too. Okay. And finally. 00:03:56.479 --> 00:03:58.960 The social media distributor. The publishing 00:03:58.960 --> 00:04:01.340 department. Yeah. A tool to automatically post 00:04:01.340 --> 00:04:03.719 your finished video. everywhere why not just 00:04:03.719 --> 00:04:06.159 connect directly to youtube or instagram because 00:04:06.159 --> 00:04:08.800 each platform has its own complex and honestly 00:04:08.800 --> 00:04:13.400 often changing api rules it's a pain a service 00:04:13.400 --> 00:04:15.759 like potato specializes in keeping up with all 00:04:15.759 --> 00:04:17.959 those changes for you so you connect your accounts 00:04:17.959 --> 00:04:20.379 youtube insta tick tock get their individual 00:04:20.379 --> 00:04:23.819 account ids plus the potato service's main api 00:04:23.819 --> 00:04:26.220 key so laying all this groundwork this upfront 00:04:26.220 --> 00:04:29.139 prep it seems critical but beyond just smoothing 00:04:29.139 --> 00:04:31.670 things out What's the bigger, maybe strategic 00:04:31.670 --> 00:04:34.290 payoff of doing all this setup so carefully? 00:04:34.550 --> 00:04:36.410 It's really about building a resilient foundation. 00:04:36.910 --> 00:04:39.129 You avoid constantly putting out fires later. 00:04:39.290 --> 00:04:41.509 It sets you up not just for smooth running now, 00:04:41.670 --> 00:04:43.529 but for scaling up and adapting down the road 00:04:43.529 --> 00:04:46.170 without rebuilding everything. All right, foundation 00:04:46.170 --> 00:04:48.430 -ly, let's talk about the engine, the idea generation 00:04:48.430 --> 00:04:51.389 module. The goal here is to get the AI to produce 00:04:51.389 --> 00:04:54.589 a single, unique video idea each time it runs. 00:04:55.019 --> 00:04:56.899 Yep. And it all starts with the ignition switch, 00:04:57.019 --> 00:04:59.660 the scheduled trigger. This node just starts 00:04:59.660 --> 00:05:01.639 the whole workflow automatically. You set the 00:05:01.639 --> 00:05:03.959 interval like every four hours. That gives you 00:05:03.959 --> 00:05:06.620 a steady flow, maybe six videos a day. And the 00:05:06.620 --> 00:05:08.860 first thing it does when it wakes up? Checks 00:05:08.860 --> 00:05:11.839 its memory. It uses a Google Sheets node to fetch 00:05:11.839 --> 00:05:15.040 all the data from that ASMR video memory spreadsheet 00:05:15.040 --> 00:05:18.040 we set up. Just pulls the list of recent creations. 00:05:18.339 --> 00:05:21.339 But that spreadsheet data isn't quite right for 00:05:21.339 --> 00:05:24.310 the AI you mentioned. Not directly, no. It's 00:05:24.310 --> 00:05:26.449 structured for a spreadsheet, not natural language. 00:05:26.829 --> 00:05:29.769 So there's a quick step using usually a couple 00:05:29.769 --> 00:05:31.730 of nodes, like an aggregate and maybe a code 00:05:31.730 --> 00:05:34.470 node, to translate that structured data into 00:05:34.470 --> 00:05:36.470 just a simple comma -separated string for the 00:05:36.470 --> 00:05:40.149 AI. Like glass pineapple, crystal dragon fruit, 00:05:40.350 --> 00:05:43.670 glass apple. Ah, I see. Just a clean list. Exactly. 00:05:43.790 --> 00:05:45.649 Makes it easy for the AI to understand what's 00:05:45.649 --> 00:05:47.629 been done recently. Okay, then the AI creative 00:05:47.629 --> 00:05:51.170 director steps in. An AI agent node using our 00:05:51.170 --> 00:05:53.480 language model. Right. And the system prompt 00:05:53.480 --> 00:05:55.300 here is crucial. It's like the job description. 00:05:55.500 --> 00:05:58.339 You tell it, select one different fruit, one 00:05:58.339 --> 00:06:00.740 not on the list, that would look visually compelling 00:06:00.740 --> 00:06:03.180 as a translucent glass object being cut. You 00:06:03.180 --> 00:06:05.639 give criteria, recognizable shape, practical 00:06:05.639 --> 00:06:07.540 to cut, elegant in glass, that kind of thing. 00:06:07.620 --> 00:06:10.420 And the user prompt. That just inserts the formatted 00:06:10.420 --> 00:06:13.220 list of recent objects we prepared. And importantly, 00:06:13.420 --> 00:06:15.680 you tell the AI to give its answer back in a 00:06:15.680 --> 00:06:18.819 specific format using structured output, usually 00:06:18.819 --> 00:06:22.500 JSON. So you get object like Glass Kiwi and caption 00:06:22.500 --> 00:06:25.560 like Glass Kiwi ASMR. Nice and tidy. So by the 00:06:25.560 --> 00:06:27.339 end of this part, we've got a fresh, unique idea 00:06:27.339 --> 00:06:29.740 and a caption all packaged up. Yep. Ready for 00:06:29.740 --> 00:06:32.220 the next stage. Now, about that memory step, 00:06:32.379 --> 00:06:35.839 feeding the AI its past work. How crucial is 00:06:35.839 --> 00:06:37.680 that really for preventing repetition? Does it 00:06:37.680 --> 00:06:40.449 make a big difference? Oh, absolutely. By explicitly 00:06:40.449 --> 00:06:44.110 showing the AI its recent history, it intelligently 00:06:44.110 --> 00:06:46.810 steers away from repeating those specific ideas. 00:06:47.029 --> 00:06:49.490 It ensures the output feels consistently fresh, 00:06:49.649 --> 00:06:52.230 not like it's stuck in a loop. Okay, now we move 00:06:52.230 --> 00:06:54.029 to the video production module. This is where 00:06:54.029 --> 00:06:56.629 we take that simple idea, like glass pineapple, 00:06:56.930 --> 00:07:00.279 and turn it into something... much more detailed, 00:07:00.360 --> 00:07:02.779 an artistic vision almost. Right. And this starts 00:07:02.779 --> 00:07:04.899 with another AI agent. Let's call it the master 00:07:04.899 --> 00:07:07.540 artisan. Its only job is to write really detailed 00:07:07.540 --> 00:07:09.420 video instructions. We're going from a simple 00:07:09.420 --> 00:07:12.300 concept to a rich... cinematic prompt and the 00:07:12.300 --> 00:07:14.839 art of the prompt is key here you mentioned totally 00:07:14.839 --> 00:07:17.600 it's not just what to show but the feel the sound 00:07:17.600 --> 00:07:20.259 the style the system prompt is super detailed 00:07:20.259 --> 00:07:23.839 optimized for a model like google vo3 9 .16 aspect 00:07:23.839 --> 00:07:27.199 ratio cinematic photorealistic static close -up 00:07:27.199 --> 00:07:30.019 dark wood board heavy steel knife wow specific 00:07:30.019 --> 00:07:32.699 yeah it even says the glass should fracture smoothly 00:07:32.699 --> 00:07:35.339 not splinter video starts right with the slicing 00:07:35.339 --> 00:07:38.240 specifics on the glass color tint no humans only 00:07:38.240 --> 00:07:41.360 hands and the sound Oh yeah, crucial for ASMR. 00:07:41.819 --> 00:07:45.480 It asks for four distinct audio layers. Crisp 00:07:45.480 --> 00:07:49.160 knife tap on glass, clean slicing sound, muted 00:07:49.160 --> 00:07:52.240 wooden knock, and a delicate glass on wood clink 00:07:52.240 --> 00:07:54.379 when the piece falls. And it emphasizes that 00:07:54.379 --> 00:07:57.139 initial knife tap sound. Yeah, even with all 00:07:57.139 --> 00:07:58.959 these specific details, I gotta admit, I still 00:07:58.959 --> 00:08:01.600 wrestle with prompt drift myself sometimes. Getting 00:08:01.600 --> 00:08:03.759 the AI to interpret exactly what you meant, getting 00:08:03.759 --> 00:08:06.639 that perfect tone. It could be tricky. You think 00:08:06.639 --> 00:08:08.379 you've nailed it, and it comes back with something 00:08:08.379 --> 00:08:11.319 slightly off. It's a constant learning process. 00:08:11.399 --> 00:08:14.660 I hear that. So this master artisan AI takes 00:08:14.660 --> 00:08:16.399 the simple object name, like glass pineapple, 00:08:16.699 --> 00:08:19.399 uses that detailed system prompt, and outputs 00:08:19.399 --> 00:08:22.439 this long descriptive paragraph. Exactly. And 00:08:22.439 --> 00:08:24.579 that paragraph becomes the direct input for the 00:08:24.579 --> 00:08:26.480 video generator service. Okay, so how do we send 00:08:26.480 --> 00:08:29.829 that order? With an HTTP request node, it's like 00:08:29.829 --> 00:08:32.090 your digital messenger. It carries that detailed 00:08:32.090 --> 00:08:34.590 order to the video production service, like fal 00:08:34.590 --> 00:08:37.870 .ai. You set the method to post, meaning create 00:08:37.870 --> 00:08:39.889 something new, and you use your video service 00:08:39.889 --> 00:08:42.570 API key to authenticate. And the body of that 00:08:42.570 --> 00:08:45.889 request, it's JSON. It includes that detailed 00:08:45.889 --> 00:08:48.509 prompt from the master artisan, specifies the 00:08:48.509 --> 00:08:52.230 aspect ratio, 9 .1 .12, and sets generate audio 00:08:52.230 --> 00:08:55.000 to true. Tells the service exactly what you want. 00:08:55.059 --> 00:08:57.039 Then comes the waiting game, right? These videos 00:08:57.039 --> 00:08:59.379 don't generate instantly. Nope. And that's where 00:08:59.379 --> 00:09:01.679 the pulling loop comes in. It's essential for 00:09:01.679 --> 00:09:04.679 tasks that take time. Think ordering coffee at 00:09:04.679 --> 00:09:07.100 a busy cafe. They give you a buzzer, right? You 00:09:07.100 --> 00:09:08.860 don't just stare at the barista. Right. You wait 00:09:08.860 --> 00:09:11.580 for the buzz. Exactly. This loop is our digital 00:09:11.580 --> 00:09:14.320 buzzer system. Conceptually, you build a small 00:09:14.320 --> 00:09:17.419 loop. First, a wait note pauses things for, say, 00:09:17.500 --> 00:09:19.659 five minutes, giving the system a head start. 00:09:20.200 --> 00:09:22.740 The response from your initial post request gives 00:09:22.740 --> 00:09:25.779 you a unique requested your order number. Then 00:09:25.779 --> 00:09:29.240 a second HTTP request node uses a get method. 00:09:29.440 --> 00:09:32.279 It calls a different status check URL from the 00:09:32.279 --> 00:09:34.220 video service, including that requested. It's 00:09:34.220 --> 00:09:36.940 basically asking, is order number 123 ready yet? 00:09:37.019 --> 00:09:39.919 A router node checks the answer. If the status 00:09:39.919 --> 00:09:42.679 is complete, great, the workflow moves on. If 00:09:42.679 --> 00:09:44.960 it's in progress, it loops back after another 00:09:44.960 --> 00:09:47.919 short, wait, maybe 30 seconds, and politely asks 00:09:47.919 --> 00:09:50.519 again, is it done yet? It just keeps checking 00:09:50.519 --> 00:09:53.600 until it gets a yes. Whoa. Imagine scaling this. 00:09:54.259 --> 00:09:56.980 Millions of videos, maybe, all patiently waiting 00:09:56.980 --> 00:09:59.940 in these loops, just being created in the background. 00:09:59.960 --> 00:10:03.120 The sheer volume of content generated automatically. 00:10:03.679 --> 00:10:06.019 That's a powerful thought. Really changes the 00:10:06.019 --> 00:10:08.139 scale of creation. It really does. Okay, so once 00:10:08.139 --> 00:10:10.720 the polling loop gets that complete status, there's 00:10:10.720 --> 00:10:12.679 usually a quick set node just before the next 00:10:12.679 --> 00:10:15.149 part. Its only job is to grab the final video 00:10:15.149 --> 00:10:17.730 URL from the polling loop's output and kind of 00:10:17.730 --> 00:10:20.029 clean things up, removing any other data we don't 00:10:20.029 --> 00:10:22.409 need anymore. That polling loop seems like the 00:10:22.409 --> 00:10:24.970 unsung hero here, ensuring things don't break 00:10:24.970 --> 00:10:26.870 just because a video takes longer than expected. 00:10:27.029 --> 00:10:28.970 What's its absolute core value, would you say? 00:10:29.230 --> 00:10:32.129 Robustness. It makes the system resilient to 00:10:32.129 --> 00:10:34.750 unpredictable delays. It ensures the factory 00:10:34.750 --> 00:10:37.190 doesn't just grind to a halt if one video takes 00:10:37.190 --> 00:10:40.129 longer, guaranteeing completion. Mid -roll sponsor 00:10:40.129 --> 00:10:43.019 read. Okay, the video is finally made. Now it's 00:10:43.019 --> 00:10:46.100 time to update our memory and share the creation. 00:10:46.379 --> 00:10:48.659 But you said update the memory first. Yes, absolutely. 00:10:48.860 --> 00:10:50.840 Before doing anything else, we record what was 00:10:50.840 --> 00:10:53.159 just made. Why delete first, though? You mentioned 00:10:53.159 --> 00:10:56.000 deleting the oldest entry. Right. We always want 00:10:56.000 --> 00:10:57.659 our memory sheet to contain the most recent, 00:10:57.740 --> 00:11:01.820 say, seven items. That gives the AI good current 00:11:01.820 --> 00:11:05.350 context for the next idea. By deleting the oldest 00:11:05.350 --> 00:11:08.309 row, like row two, if row one is headers, and 00:11:08.309 --> 00:11:10.210 then adding the new one at the bottom, we create 00:11:10.210 --> 00:11:13.210 a rolling list. It's a classic first in, first 00:11:13.210 --> 00:11:16.549 out, or FIFO system, like a queue. Ah, okay, 00:11:16.690 --> 00:11:18.889 like a conveyor belt of ideas. So that involves 00:11:18.889 --> 00:11:21.450 two Google Sheets nodes. Typically, yeah. One 00:11:21.450 --> 00:11:23.509 set to delete row two, and then another one right 00:11:23.509 --> 00:11:26.149 after, set to append row, inserting the new object 00:11:26.149 --> 00:11:27.990 and the final video roll we got from the production 00:11:27.990 --> 00:11:30.289 stage. Got it. Memory updated, now publishing. 00:11:30.570 --> 00:11:34.149 Yep, the final step, sending our video out. And 00:11:34.149 --> 00:11:36.389 a smart move here, especially at first, is to 00:11:36.389 --> 00:11:40.370 test using private or unlisted visibility. Just 00:11:40.370 --> 00:11:42.809 to check, the post looks right on each platform 00:11:42.809 --> 00:11:45.470 before going public. Makes sense. How's that 00:11:45.470 --> 00:11:47.830 done automatically? Usually with four separate 00:11:47.830 --> 00:11:51.330 HTTP request nodes. One uploads the video file 00:11:51.330 --> 00:11:53.710 or provides the URL to your distribution service, 00:11:53.850 --> 00:11:57.129 like Plotato. Then you have three more nodes 00:11:57.129 --> 00:12:00.070 branching off from that upload node, one each 00:12:00.070 --> 00:12:03.909 for YouTube, Instagram, and TikTok. Connecting 00:12:03.909 --> 00:12:05.870 them like this means they can run in parallel 00:12:05.870 --> 00:12:09.730 all at the same time, which is faster. All four 00:12:09.730 --> 00:12:11.970 use your social media distribution service's 00:12:11.970 --> 00:12:14.929 main API key. Each of the final three nodes, 00:12:15.110 --> 00:12:17.350 YouTube, Insta, TikTok, has a slightly different 00:12:17.350 --> 00:12:20.929 JSON body. You insert the video URL, the caption 00:12:20.929 --> 00:12:23.370 we generated way back in the idea stage, and 00:12:23.370 --> 00:12:25.370 the specific account ID for that platform that 00:12:25.370 --> 00:12:28.730 you saved during setup. You just swish it on. 00:12:28.809 --> 00:12:30.789 Pretty much. You activate the workflow. Your 00:12:30.789 --> 00:12:33.350 autonomous ASMR video factory is now open for 00:12:33.350 --> 00:12:35.830 business. So that detail about updating the memory 00:12:35.830 --> 00:12:38.370 before distributing, why is that sequence so 00:12:38.370 --> 00:12:40.889 vital? What's the core reason? It guarantees 00:12:40.889 --> 00:12:44.090 the system always knows its absolute latest creation 00:12:44.090 --> 00:12:46.710 before potentially starting the next cycle. It 00:12:46.710 --> 00:12:49.870 closes the loop, preventing any chance of immediately 00:12:49.870 --> 00:12:52.450 repeating the same idea. OK, but before you set 00:12:52.450 --> 00:12:55.519 this thing to run 24 -7. Churning out glass fruit 00:12:55.519 --> 00:12:58.039 videos. We need to talk economics, right? This 00:12:58.039 --> 00:13:01.200 isn't free. Definitely not. Access to high end 00:13:01.200 --> 00:13:04.299 AI video models costs money. The source estimated, 00:13:04.500 --> 00:13:06.960 let's say, roughly six dollars for a high quality 00:13:06.960 --> 00:13:09.700 eight second video with generated audio. Could 00:13:09.700 --> 00:13:12.279 be more, could be less. But let's use that. Right. 00:13:12.379 --> 00:13:14.620 So if your workflow runs six times a day, like 00:13:14.620 --> 00:13:16.899 we discussed earlier, your daily cost is around 00:13:16.899 --> 00:13:19.870 thirty six dollars. Which translates to. what, 00:13:19.909 --> 00:13:22.570 about $1 ,080 a month? Yeah. So yeah, it's important 00:13:22.570 --> 00:13:24.450 to frame this correctly. This isn't some kind 00:13:24.450 --> 00:13:26.789 of get rich quick button. Exactly. It's a strategic 00:13:26.789 --> 00:13:28.830 investment. You're funding a high volume content 00:13:28.830 --> 00:13:30.990 production machine. The thinking is really a 00:13:30.990 --> 00:13:33.429 numbers game. More volume equals more chances 00:13:33.429 --> 00:13:35.529 for a video to hit the algorithm just right and 00:13:35.529 --> 00:13:38.110 go viral. And the potential value of one viral 00:13:38.110 --> 00:13:41.029 hit. That can be huge, right? Millions of views 00:13:41.029 --> 00:13:43.049 could mean significant ad revenue, maybe brand 00:13:43.049 --> 00:13:45.549 sponsorships, a big jump in followers you can 00:13:45.549 --> 00:13:48.970 monetize later. For sure. One really successful 00:13:48.970 --> 00:13:51.690 video could potentially cover months of operating 00:13:51.690 --> 00:13:54.549 costs. And it gives you a competitive edge, too. 00:13:54.730 --> 00:13:58.049 This content style is popular, but very few creators 00:13:58.049 --> 00:14:00.929 can pump it out at this scale manually. Automation 00:14:00.929 --> 00:14:04.149 allows a volume that's just physically impossible 00:14:04.149 --> 00:14:06.919 for a person. Right. And beyond ads, you can 00:14:06.919 --> 00:14:09.279 monetize with affiliate links, maybe merchandise, 00:14:09.580 --> 00:14:12.179 or even use the channel as a sort of portfolio 00:14:12.179 --> 00:14:14.899 for other video services. So looking at that, 00:14:15.000 --> 00:14:18.860 say, $1 ,000 a month cost, what's the fundamental 00:14:18.860 --> 00:14:22.419 shift in mindset needed to see that as an investment, 00:14:22.559 --> 00:14:25.279 not just an expense? It's about funding a scalable 00:14:25.279 --> 00:14:27.960 content engine. You're leveraging volume to play 00:14:27.960 --> 00:14:30.059 the algorithmic game and opening up multiple 00:14:30.059 --> 00:14:32.279 ways to get a return, not just focusing on the 00:14:32.279 --> 00:14:34.679 per video cost. This whole workflow, this glass 00:14:34.679 --> 00:14:37.740 fruit video factory, it's fascinating on its 00:14:37.740 --> 00:14:39.740 own, but it feels like it's more than that. It's 00:14:39.740 --> 00:14:41.820 like a perfect case study in smart automation 00:14:41.820 --> 00:14:44.000 design, isn't it? Absolutely. The concepts here, 00:14:44.019 --> 00:14:45.980 you can apply them to almost any automated system 00:14:45.980 --> 00:14:48.840 you can think of. Like what? What are the core 00:14:48.840 --> 00:14:51.919 principles we can take away? Okay, first, context 00:14:51.919 --> 00:14:55.360 is king. A smart system remembers its past actions. 00:14:55.539 --> 00:14:58.000 That simple Google sheet acting as short -term 00:14:58.000 --> 00:15:00.899 memory. It stops the AI making repetitive choices, 00:15:01.100 --> 00:15:04.679 keeps things fresh, embrace uncertainty. The 00:15:04.679 --> 00:15:08.039 real world, especially AI generation, is an instant. 00:15:08.220 --> 00:15:11.240 It takes unpredictable time. The system doesn't 00:15:11.240 --> 00:15:13.799 fail. It uses that professional polling loop 00:15:13.799 --> 00:15:17.539 to patiently check status. Makes it robust. Third, 00:15:17.720 --> 00:15:20.960 work in parallel. Time's valuable. Publishing 00:15:20.960 --> 00:15:23.059 to YouTube, Instagram, TikTok simultaneously 00:15:23.059 --> 00:15:26.059 at the end, that makes the final step as fast 00:15:26.059 --> 00:15:28.460 and efficient as possible. Designed for autonomy. 00:15:28.659 --> 00:15:31.080 From the scheduled trigger starting at all to 00:15:31.080 --> 00:15:33.240 the automatic memory updates the system manages 00:15:33.240 --> 00:15:36.120 itself. The goal of great automation is minimal 00:15:36.120 --> 00:15:39.679 ongoing human babysitting for core tasks. Let 00:15:39.679 --> 00:15:42.159 it run itself. And finally, modularity. Each 00:15:42.159 --> 00:15:45.379 part, idea gen, video prod, distribution is its 00:15:45.379 --> 00:15:47.500 own module. That's key. You can swap out one 00:15:47.500 --> 00:15:49.720 part, like maybe try a new video AI service later 00:15:49.720 --> 00:15:51.480 without having to rebuild the entire factory. 00:15:51.639 --> 00:15:53.700 It really is an extraordinary time we're living 00:15:53.700 --> 00:15:56.700 in. The ability to just have an idea, have an 00:15:56.700 --> 00:16:00.039 AI create. professional level video, distribute 00:16:00.039 --> 00:16:02.860 it globally, and the system manages itself. That 00:16:02.860 --> 00:16:04.919 felt like pure science fiction just a few years 00:16:04.919 --> 00:16:07.559 back. For sure. And this ASMR factory is just 00:16:07.559 --> 00:16:10.200 one example. You could adapt this exact structure, 00:16:10.259 --> 00:16:13.059 right? Automated news summaries, little history 00:16:13.059 --> 00:16:16.000 fact shorts, product demo clips. The underlying 00:16:16.000 --> 00:16:18.860 pattern holds. Yeah. The idea engine, the prompt 00:16:18.860 --> 00:16:21.159 artisan, the production team, the distribution 00:16:21.159 --> 00:16:24.600 manager. Those roles remain the same, just the 00:16:24.600 --> 00:16:26.580 specifics change. And it's important to remember. 00:16:27.159 --> 00:16:30.059 Automation like this, it's not really about replacing 00:16:30.059 --> 00:16:32.320 human creativity, is it? Not at all. It's about 00:16:32.320 --> 00:16:35.720 amplifying it. It's a tool, a powerful one, that 00:16:35.720 --> 00:16:38.360 takes away the tedious, repetitive, time -sucking 00:16:38.360 --> 00:16:40.700 parts of making stuff. Which frees you up. Frees 00:16:40.700 --> 00:16:43.559 you up to focus on what humans do best. The strategy, 00:16:43.659 --> 00:16:45.460 the analysis, coming up with the next brilliant 00:16:45.460 --> 00:16:47.379 idea in the first place. You basically have the 00:16:47.379 --> 00:16:50.080 whole blueprint now. So, stepping back from the 00:16:50.080 --> 00:16:53.019 build itself, what's the ultimate takeaway here 00:16:53.019 --> 00:16:55.279 about automation and human creativity working 00:16:55.279 --> 00:16:58.080 together? Automation handles the drudgery, the 00:16:58.080 --> 00:17:01.679 rote tasks. This elevates the human role to strategist, 00:17:01.700 --> 00:17:04.240 innovator, focusing on higher level thinking, 00:17:04.359 --> 00:17:07.119 and genuine creative breakthroughs. So recapping 00:17:07.119 --> 00:17:10.559 the big picture. This AI video factory deep dive 00:17:10.559 --> 00:17:13.619 really shows how you can automate complex content 00:17:13.619 --> 00:17:16.579 creation truly end to end. Yeah, it uses connected 00:17:16.579 --> 00:17:19.680 services, these smart AI agents and some clever 00:17:19.680 --> 00:17:22.619 logic like loops to handle pretty much everything 00:17:22.619 --> 00:17:24.920 automatically. But it's definitely not just about 00:17:24.920 --> 00:17:28.359 ASMR videos. This is a powerful, adaptable blueprint 00:17:28.359 --> 00:17:31.000 for almost any kind of scalable content strategy. 00:17:31.339 --> 00:17:34.339 It really highlights the power of intelligent 00:17:34.339 --> 00:17:36.980 system design and thinking strategically about. 00:17:37.640 --> 00:17:39.980 investing in these AI tools. So here's a thought 00:17:39.980 --> 00:17:42.670 for you, the listener, to maybe chew on. What 00:17:42.670 --> 00:17:45.069 repetitive creative task in your own life or 00:17:45.069 --> 00:17:47.390 maybe in your work could be fundamentally transformed 00:17:47.390 --> 00:17:49.930 if you applied these same automation principles? 00:17:50.130 --> 00:17:52.230 The tools are definitely here. The blueprint's 00:17:52.230 --> 00:17:54.369 pretty clear now. The next step really, well, 00:17:54.589 --> 00:17:56.390 that's always up to you, isn't it? We really 00:17:56.390 --> 00:17:59.130 hope this deep dive sparked some aha moments, 00:17:59.329 --> 00:18:01.509 maybe inspired you to explore what's becoming 00:18:01.509 --> 00:18:04.069 possible with intelligent automation. Yeah, absolutely. 00:18:04.329 --> 00:18:06.630 Until next time, keep exploring, keep learning. 00:18:07.029 --> 00:18:07.890 Outiro music.