WEBVTT 00:00:00.000 --> 00:00:02.700 You know, high quality cinematic AI video. It 00:00:02.700 --> 00:00:05.200 used to hit your wallet at about, what, a dollar 00:00:05.200 --> 00:00:07.860 for every 10 seconds? Yeah, roughly. That was 00:00:07.860 --> 00:00:09.359 kind of the price for anything professional, 00:00:09.560 --> 00:00:12.320 you know, watermark free. Right. But that landscape 00:00:12.320 --> 00:00:14.019 just completely fractured. We've actually found 00:00:14.019 --> 00:00:16.460 a way to get that exact same professional output. 00:00:16.640 --> 00:00:18.800 Well, through automation, basically. And it comes 00:00:18.800 --> 00:00:21.320 down to just 15 cents for 10 seconds. 15 cents. 00:00:21.359 --> 00:00:24.199 Yeah. That's a six -fold cost advantage. We're 00:00:24.199 --> 00:00:25.940 not just talking iteration here. This is like, 00:00:25.980 --> 00:00:29.929 it obliterates the old price model. Wow. Okay, 00:00:30.010 --> 00:00:33.289 welcome to the deep dive. Our mission today is 00:00:33.289 --> 00:00:35.450 pretty straightforward, I think. We are going 00:00:35.450 --> 00:00:38.009 to tear apart the technical blueprint you need 00:00:38.009 --> 00:00:40.890 to build an automated, scalable video content 00:00:40.890 --> 00:00:44.340 machine. Exactly. Our sources lay out the precise 00:00:44.340 --> 00:00:47.520 steps for combining Sora to, you know, the state 00:00:47.520 --> 00:00:49.880 of the art AI video generator with N8n, which 00:00:49.880 --> 00:00:52.539 is this really powerful no code platform that 00:00:52.539 --> 00:00:54.920 kind of glues everything together. So if you 00:00:54.920 --> 00:00:56.840 want to build a content factory, this is pretty 00:00:56.840 --> 00:00:59.000 much your operating manual. So we'll start by 00:00:59.000 --> 00:01:01.600 digging into the business case. Right. Why this 00:01:01.600 --> 00:01:04.280 sixfold cost reduction is so important. Yep. 00:01:04.340 --> 00:01:06.340 And then we'll get into the nuts and bolts, the 00:01:06.340 --> 00:01:09.140 fundamental two step API conversation you need 00:01:09.140 --> 00:01:12.810 for the automation side. OK. And finally. Finally, 00:01:12.849 --> 00:01:15.370 we'll explore the really cool stuff, the advanced 00:01:15.370 --> 00:01:17.730 applications like creating consistent storyboards 00:01:17.730 --> 00:01:22.109 and mastering the art of the AI optimized prompt. 00:01:22.370 --> 00:01:23.950 You know, when Sora 2 first came out, everyone 00:01:23.950 --> 00:01:26.090 was focused on the cool videos, the high fidelity 00:01:26.090 --> 00:01:28.989 stuff. Yeah, that candy. Right. But the real 00:01:28.989 --> 00:01:30.769 breakthrough, it seems, is in its professional 00:01:30.769 --> 00:01:33.150 application. This feels like a tool built for 00:01:33.150 --> 00:01:36.849 business scale, not just, you know. funny experiments. 00:01:37.269 --> 00:01:39.569 Precisely. We are seeing really for the very 00:01:39.569 --> 00:01:42.170 first time professional grade video production 00:01:42.170 --> 00:01:44.890 becoming automated and maybe most critically, 00:01:45.030 --> 00:01:48.569 just dirt cheap. And that absolutely disrupts 00:01:48.569 --> 00:01:50.590 the old way of doing things, the old economic 00:01:50.590 --> 00:01:53.450 model for video. So what does that actually look 00:01:53.450 --> 00:01:55.590 like for, say, a business owner or a marketer 00:01:55.590 --> 00:01:58.189 on the ground? OK, think about volume, just sheer 00:01:58.189 --> 00:02:01.530 volume. You can now create endless variations 00:02:01.530 --> 00:02:04.530 of high converting UGC. That's user generated 00:02:04.530 --> 00:02:07.310 content style ads, you know, for TikTok, Instagram, 00:02:07.390 --> 00:02:10.310 wherever. OK. Or you could generate hyper realistic 00:02:10.310 --> 00:02:12.909 product demos without ever stepping into a studio. 00:02:12.990 --> 00:02:16.409 No actor, no film permit needed. That used to 00:02:16.409 --> 00:02:18.689 be a huge hurdle, right? Especially for smaller 00:02:18.689 --> 00:02:20.750 businesses. It really was. I mean, the cost used 00:02:20.750 --> 00:02:23.310 to involve paying crews, dealing with insurance, 00:02:23.710 --> 00:02:26.430 hoping the weather holds up. The main cost is 00:02:26.430 --> 00:02:29.770 literally a few cents for AI processing. And 00:02:29.770 --> 00:02:31.669 it happens almost instantly from just a well 00:02:31.669 --> 00:02:34.069 -written text prompt. That kind of scale just 00:02:34.069 --> 00:02:36.150 fundamentally changes how brands can approach 00:02:36.150 --> 00:02:38.650 content. OK. So the economic upside is clear. 00:02:38.810 --> 00:02:42.849 If that old cost barrier is just gone, what's 00:02:42.849 --> 00:02:45.810 the single biggest area you think we'll see immediate 00:02:45.810 --> 00:02:48.569 like instant growth? High volume social media 00:02:48.569 --> 00:02:51.389 ads and endless A -B testing will explode instantly. 00:02:51.669 --> 00:02:53.770 OK. That covers the why. Let's talk money then. 00:02:54.030 --> 00:02:56.370 How do we actually get that 15 cents for 10 seconds 00:02:56.370 --> 00:02:58.229 price? Sounds like it comes down to the platform 00:02:58.229 --> 00:03:00.949 you use. Exactly. We found this 6x cost advantage 00:03:00.949 --> 00:03:03.550 comes from routing Sora 2, processing through 00:03:03.550 --> 00:03:06.789 something called Key AI, K -I -E dot AI. Yeah, 00:03:06.830 --> 00:03:09.830 Key AI. It's an AI model aggregator, and it's 00:03:09.830 --> 00:03:12.669 a total game changer because it lets you sidestep 00:03:12.669 --> 00:03:16.509 the standard OpenAI API costs. Which are much 00:03:16.509 --> 00:03:18.610 higher. Oh, yeah. They clock in around 10 cents 00:03:18.610 --> 00:03:21.810 per second. So the numbers really force you into 00:03:21.810 --> 00:03:24.090 a strategic choice here. They really do. Think 00:03:24.090 --> 00:03:27.530 about it. A standard workflow, maybe 100 videos 00:03:27.530 --> 00:03:30.129 a month, that's a decent volume for content. 00:03:30.250 --> 00:03:33.409 It used to cost you, what, $1 ,000. Now that 00:03:33.409 --> 00:03:37.590 same output is $150. Wow. We are literally talking 00:03:37.590 --> 00:03:41.050 about saving over $10 ,000 a year just by choosing 00:03:41.050 --> 00:03:43.210 the right API access point. But that kind of 00:03:43.210 --> 00:03:45.409 saving makes something like A -B testing completely 00:03:45.409 --> 00:03:47.949 different. Totally. It moves it from this thing 00:03:47.949 --> 00:03:50.110 you wish you could do to just daily operations. 00:03:50.509 --> 00:03:52.370 You know, I still wrestle with prompt drift myself 00:03:52.370 --> 00:03:55.050 sometimes. Yeah, like getting the AI to keep 00:03:55.050 --> 00:03:57.069 a character looking the same across different 00:03:57.069 --> 00:04:00.090 shots. It's tricky. But honestly, the sheer cost 00:04:00.090 --> 00:04:03.110 barrier used to stop me from even trying that 00:04:03.110 --> 00:04:05.430 many variations. Now there's just so much more. 00:04:05.550 --> 00:04:08.069 room to experiment okay so for someone listening 00:04:08.069 --> 00:04:10.090 who wants to just jump in maybe this afternoon 00:04:10.090 --> 00:04:13.969 what are the immediate steps the tactical stuff 00:04:13.969 --> 00:04:17.470 okay it's a pretty clean three -step setup first 00:04:17.470 --> 00:04:20.319 you got to create your key ai account Add some 00:04:20.319 --> 00:04:23.040 credits, seriously. Like, $5 gives you a ton 00:04:23.040 --> 00:04:25.100 of runway to test things out. Just $5. Okay. 00:04:25.220 --> 00:04:27.860 Yeah. Second, generate that API key and save 00:04:27.860 --> 00:04:29.740 it somewhere secure. That's your credential. 00:04:29.839 --> 00:04:32.199 That's your power, basically. Right. And the 00:04:32.199 --> 00:04:35.079 third step. Focus on the standard Sora 2 model 00:04:35.079 --> 00:04:38.199 within Key AI. Our sources suggest that model 00:04:38.199 --> 00:04:40.680 gives you the perfect blend of quality and speed 00:04:40.680 --> 00:04:43.160 for automation, and it keeps that cost advantage. 00:04:43.660 --> 00:04:46.360 Mm -hmm. The quality's great, but there's another 00:04:46.360 --> 00:04:48.560 reason to use an aggregator like Key AI, isn't 00:04:48.560 --> 00:04:51.459 there? Beyond just the 6x cost saving. If the 00:04:51.459 --> 00:04:54.079 6x cost savings weren't enough, what does using 00:04:54.079 --> 00:04:56.439 key AI unlock that the standard direct model 00:04:56.439 --> 00:04:59.319 doesn't? The superior quality and ability to 00:04:59.319 --> 00:05:02.139 remove watermarks unlock commercial use. Okay, 00:05:02.180 --> 00:05:06.000 so moving from cost to mechanics. To automate 00:05:06.000 --> 00:05:09.459 any AI model reliably, you really have to understand 00:05:09.459 --> 00:05:12.980 the basic conversation your automation... platform 00:05:12.980 --> 00:05:15.160 in this case, and it needs to help with the API. 00:05:15.819 --> 00:05:19.620 It's never just a single command. Right. We found 00:05:19.620 --> 00:05:22.300 you basically have to talk to the AI twice. Our 00:05:22.300 --> 00:05:24.480 sources call this the fundamental to request 00:05:24.480 --> 00:05:26.939 pattern for Sora 2. It's kind of like a simple 00:05:26.939 --> 00:05:29.540 dance. You place the order. Then you check on 00:05:29.540 --> 00:05:31.420 the delivery. Exactly. Request one is the go 00:05:31.420 --> 00:05:33.360 command. You send your detailed text prompt, 00:05:33.519 --> 00:05:35.899 your technical parameters, all that. The API 00:05:35.899 --> 00:05:38.160 immediately fires back with a task ID. Okay. 00:05:38.279 --> 00:05:40.579 Think of that task ID as your digital receipt. 00:05:40.779 --> 00:05:42.959 It's proof your video is being rendered somewhere. 00:05:43.259 --> 00:05:45.519 And request two is the crucial follow -up, the 00:05:45.519 --> 00:05:47.819 check -in. That's right. You then have to periodically 00:05:47.819 --> 00:05:50.980 pull, basically, check in with the API using 00:05:50.980 --> 00:05:53.300 that task ID. You keep checking until the response 00:05:53.300 --> 00:05:55.180 state finally says success. And then you get 00:05:55.180 --> 00:05:57.560 the video. Then you get the crucial download 00:05:57.560 --> 00:05:59.879 link. That whole start a poll, retrieve sequence, 00:06:00.220 --> 00:06:03.600 that's the bedrock. Honestly, every advanced 00:06:03.600 --> 00:06:05.459 workflow we're going to talk about relies on 00:06:05.459 --> 00:06:07.800 getting this right. So we build this in NANN 00:06:07.800 --> 00:06:10.360 using the HTTP request node. Yeah. That node 00:06:10.360 --> 00:06:12.959 is like your telephone to the outside world to 00:06:12.959 --> 00:06:15.339 the API. Right. And to keep things scalable and 00:06:15.339 --> 00:06:17.399 secure, you need to handle the authentication 00:06:17.399 --> 00:06:19.920 part efficiently. Oh, absolutely, yes. please 00:06:19.920 --> 00:06:22.040 don't waste time manually pasting your secret 00:06:22.040 --> 00:06:25.019 key everywhere. Set up a reusable key AI credential 00:06:25.019 --> 00:06:28.060 in NN. How does that work? You input your API 00:06:28.060 --> 00:06:31.019 key just once, making sure you include the prefix 00:06:31.019 --> 00:06:33.680 bearer, B -E -A -R -E -R space before the key 00:06:33.680 --> 00:06:37.160 itself. Save it. Then that credential can be 00:06:37.160 --> 00:06:39.600 reused across, I don't know, a thousand different 00:06:39.600 --> 00:06:41.220 automation nodes if you need. Way more efficient. 00:06:41.579 --> 00:06:43.360 Way more secure. Okay. That makes sense. And 00:06:43.360 --> 00:06:45.819 the actual order form, the JSON body you send 00:06:45.819 --> 00:06:48.180 in request one, that has to be perfect, right? 00:06:48.279 --> 00:06:50.819 What fields are absolutely essential. Yeah. The 00:06:50.819 --> 00:06:53.759 required fields are surprisingly clean, actually. 00:06:53.800 --> 00:06:56.540 You need the main prompt, obviously. The desired 00:06:56.540 --> 00:06:59.300 aspect ratio. The end frames, which basically 00:06:59.300 --> 00:07:01.600 controls the video length. And this critical 00:07:01.600 --> 00:07:05.230 one, remove watermark. True. Ah, there it is 00:07:05.230 --> 00:07:07.670 again. Yep. That's the commercial power the aggregator 00:07:07.670 --> 00:07:09.529 gives you. Okay. So if that video generation 00:07:09.529 --> 00:07:13.050 takes, let's say, five minutes. Yeah. Which isn't 00:07:13.050 --> 00:07:16.509 unusual for complex stuff. Yeah. How do we automate 00:07:16.509 --> 00:07:20.449 that? Is it done yet? Without just wasting time 00:07:20.449 --> 00:07:22.790 or hammering their server. We build a reliable 00:07:22.790 --> 00:07:25.350 polling loop that checks status repeatedly until 00:07:25.350 --> 00:07:27.389 success. Right. Because if we had to manually 00:07:27.389 --> 00:07:29.649 check. The whole factory idea would just grind 00:07:29.649 --> 00:07:32.189 to a halt. So we bypass that with this polling 00:07:32.189 --> 00:07:34.269 weight loop. This is where tools like the IF 00:07:34.269 --> 00:07:36.889 node or the switch node in NANA really shine. 00:07:37.089 --> 00:07:39.769 Exactly. The logic is actually quite elegant. 00:07:39.870 --> 00:07:42.569 When you poll and check the status, if the result 00:07:42.569 --> 00:07:45.189 comes back, success. great you continue the workflow 00:07:45.189 --> 00:07:48.170 download the file done but if it's not done if 00:07:48.170 --> 00:07:50.589 the state is still generating the workflow just 00:07:50.589 --> 00:07:52.790 loops back to a wait node you can set it for 00:07:52.790 --> 00:07:55.850 say 30 seconds maybe less and then it tries the 00:07:55.850 --> 00:07:59.389 status check again ah so you get the video basically 00:07:59.389 --> 00:08:01.550 the instant it's ready but your workflow never 00:08:01.550 --> 00:08:03.709 times out or gets stuck while it's rendering 00:08:03.709 --> 00:08:07.189 super efficient nice let's talk about some of 00:08:07.189 --> 00:08:09.110 the high value features this kind of automated 00:08:09.110 --> 00:08:11.829 factory unlocks first one that jumps out is image 00:08:11.829 --> 00:08:15.740 to video This feels like kind of the gold standard 00:08:15.740 --> 00:08:18.449 for e -commerce and product marketing now. It 00:08:18.449 --> 00:08:21.050 truly is. It's ideal for generating those really 00:08:21.050 --> 00:08:24.750 high converting UGC ads where the actual product, 00:08:24.810 --> 00:08:27.569 let's say it's a bottle of Clarity Curls hair 00:08:27.569 --> 00:08:30.350 cream, stays perfectly crisp and accurate because 00:08:30.350 --> 00:08:33.409 you supplied the source image. But the scene, 00:08:33.549 --> 00:08:36.350 the actor, the lighting, all that stuff is generated 00:08:36.350 --> 00:08:38.389 authentically around your product image. It's 00:08:38.389 --> 00:08:40.990 powerful. But there's a really important safety 00:08:40.990 --> 00:08:42.730 restriction there, though, isn't there? Absolutely 00:08:42.730 --> 00:08:45.129 critical distinction. To prevent misuse, you 00:08:45.129 --> 00:08:47.669 cannot use a photo of a realistic person. as 00:08:47.669 --> 00:08:50.210 the source image. That's a hard block. You must 00:08:50.210 --> 00:08:52.070 use the image of the product and then you describe 00:08:52.070 --> 00:08:54.389 the person you want to see holding it or interacting 00:08:54.389 --> 00:08:57.190 with it in the text prompt itself. Got it. Good 00:08:57.190 --> 00:09:00.450 to know. Another powerful feature is video cameos. 00:09:00.570 --> 00:09:03.549 You can actually include the unique public profile 00:09:03.549 --> 00:09:06.409 name of a famous person. The example given is 00:09:06.409 --> 00:09:09.440 Sam Altman's Sama right in your prompt. Yeah. 00:09:09.559 --> 00:09:12.740 And this flips into a huge entrepreneurial opportunity. 00:09:13.120 --> 00:09:15.559 You should set up your own Cameo profile on whatever 00:09:15.559 --> 00:09:18.220 platform supports it. Or yourself. Yeah. So you 00:09:18.220 --> 00:09:20.820 can then generate unlimited, consistent AI videos 00:09:20.820 --> 00:09:24.139 of you, or at least your digital likeness, for 00:09:24.139 --> 00:09:26.860 automated personal branding. Suddenly, you can 00:09:26.860 --> 00:09:29.059 scale your own content presence without ever 00:09:29.059 --> 00:09:31.720 needing a camera crew again. It's kind of wild. 00:09:31.879 --> 00:09:34.399 That is wild. Okay. Maybe the ultimate scaling 00:09:34.399 --> 00:09:37.299 feature, especially for telling stories, is storyboards. 00:09:37.710 --> 00:09:39.909 Creating a single video with multiple scenes, 00:09:40.190 --> 00:09:42.710 three or more shots, but with a guaranteed consistent 00:09:42.710 --> 00:09:45.649 character across all of them. Whoa, yeah. Just 00:09:45.649 --> 00:09:48.570 imagine scaling a content campaign where your 00:09:48.570 --> 00:09:51.669 main brand avatar or character is perfectly consistent 00:09:51.669 --> 00:09:53.990 across, I don't know, a billion queries a year. 00:09:54.110 --> 00:09:56.289 A billion. It's absolutely mind -bending to think 00:09:56.289 --> 00:09:59.289 about that level of consistency at scale. The 00:09:59.289 --> 00:10:02.470 only caveat with storyboards is just time. They 00:10:02.470 --> 00:10:04.529 take significantly longer to process. You might 00:10:04.529 --> 00:10:07.190 be looking at 8 to 12 minutes, maybe more, per 00:10:07.190 --> 00:10:10.230 video. Okay, something to factor in. So once 00:10:10.230 --> 00:10:12.190 you've got a workflow like this running smoothly, 00:10:12.409 --> 00:10:15.889 maybe churning out hundreds of videos, what's 00:10:15.889 --> 00:10:20.289 the single most common, maybe mundane, technical 00:10:20.289 --> 00:10:22.950 problem that's likely to just break the whole 00:10:22.950 --> 00:10:25.519 factory? Simple characters like new lines or 00:10:25.519 --> 00:10:28.000 quotation marks will break the API's required 00:10:28.000 --> 00:10:30.600 JSON format. Right, because the entire quality 00:10:30.600 --> 00:10:32.919 of what comes out depends so much on the input. 00:10:33.100 --> 00:10:35.519 It feels like the single most valuable skill 00:10:35.519 --> 00:10:38.659 in this entire stack is still prompt engineering. 00:10:39.179 --> 00:10:41.399 Your text prompt really needs to function like 00:10:41.399 --> 00:10:43.679 a super detailed shot list for a cinematographer. 00:10:43.820 --> 00:10:46.639 Totally. It really needs to cover four distinct 00:10:46.639 --> 00:10:48.379 components to be truly effective. You need the 00:10:48.379 --> 00:10:50.179 main subject, right, and what they're doing, 00:10:50.200 --> 00:10:52.399 their movement. Okay. Then the setting and environment. 00:10:52.970 --> 00:10:55.250 including details like the lighting. Third, the 00:10:55.250 --> 00:10:58.250 camera style, what lens angle? Is it moving like 00:10:58.250 --> 00:11:00.690 a dolly shot or a gimbal shot? Right, the technical 00:11:00.690 --> 00:11:03.230 camera details. Exactly. And finally, the overall 00:11:03.230 --> 00:11:05.309 technical direction, which kind of covers mood, 00:11:05.490 --> 00:11:08.049 color grading, that sort of aesthetic feel. You 00:11:08.049 --> 00:11:10.409 know, a two -word prompt gets you a pretty static 00:11:10.409 --> 00:11:13.350 clip. A 150 -word cinematic prompt, that can 00:11:13.350 --> 00:11:16.169 give you a masterpiece. And if you're not, maybe... 00:11:16.350 --> 00:11:18.350 a naturally gifted cinematic writer yourself, 00:11:18.750 --> 00:11:20.889 there's that AI inception trick you mentioned 00:11:20.889 --> 00:11:23.809 earlier, using another AI agent to write the 00:11:23.809 --> 00:11:26.309 prompt for you. Yeah. Give that other AI a really 00:11:26.309 --> 00:11:28.850 detailed system prompt about what makes a good 00:11:28.850 --> 00:11:31.029 Sora prompt. Then you give it your simple concept. 00:11:31.110 --> 00:11:33.789 Like your raw input might just be professor at 00:11:33.789 --> 00:11:36.350 a blackboard, but the AI optimized prompt it 00:11:36.350 --> 00:11:38.909 generates comes back with, you know, medium shot, 00:11:39.190 --> 00:11:42.149 24 millimeter gimbal movement, smoothly dollies 00:11:42.149 --> 00:11:45.090 in towards professor, ambient classroom lighting. 00:11:45.639 --> 00:11:48.120 soft golden hour diffusion through windows, all 00:11:48.120 --> 00:11:50.740 that detail. The difference in the final video 00:11:50.740 --> 00:11:53.480 output between those two prompts is just monumental, 00:11:53.799 --> 00:11:56.419 truly. Okay, finally, we need a quick troubleshooting 00:11:56.419 --> 00:11:58.879 checklist. You mentioned the number one cause 00:11:58.879 --> 00:12:01.379 of failure in these automated workflows is often 00:12:01.379 --> 00:12:04.139 JSON errors. It always seems to come back to 00:12:04.139 --> 00:12:07.039 the format, doesn't it? Those AI -generated prompts, 00:12:07.279 --> 00:12:09.980 even good ones, often have sneaky little characters 00:12:09.980 --> 00:12:12.940 hidden in them, like new line characters or extra 00:12:12.940 --> 00:12:15.240 double quotation marks. And those break the API 00:12:15.240 --> 00:12:18.220 call. Completely. They break the rigid JSON structure 00:12:18.220 --> 00:12:22.120 the API needs. So you absolutely must clean that 00:12:22.120 --> 00:12:24.159 prompt output before you send it to the Sora 00:12:24.159 --> 00:12:27.659 API via key AI. Use cleanup expressions right 00:12:27.659 --> 00:12:30.840 in N8n to strip those characters out. It's easy 00:12:30.840 --> 00:12:33.629 once you know to look for it. Good tip. And what 00:12:33.629 --> 00:12:37.250 about error codes? If the workflow fails, what 00:12:37.250 --> 00:12:39.250 should people look for? Okay, two common ones. 00:12:39.649 --> 00:12:42.429 If you get a 500 internal server error, that 00:12:42.429 --> 00:12:44.710 generally means their server is overloaded. Or 00:12:44.710 --> 00:12:46.809 maybe you accidentally triggered a content restriction 00:12:46.809 --> 00:12:49.669 filter. What's the fix? Usually simple. Just 00:12:49.669 --> 00:12:51.629 wait five or ten minutes and retry the workflow. 00:12:51.730 --> 00:12:53.710 It often clears itself up. And the other one? 00:12:53.789 --> 00:12:56.389 If you see a 402 payment required code, well, 00:12:56.429 --> 00:12:59.159 that's even simpler. Let me guess. Yep. You ran 00:12:59.159 --> 00:13:01.340 out of credits in your key AI account. Easiest 00:13:01.340 --> 00:13:03.860 fix on the list. Just top up your account balance 00:13:03.860 --> 00:13:06.100 and hit run again. Hashtag tag tag conclusion 00:13:06.100 --> 00:13:09.100 and outro. So I think we've really established 00:13:09.100 --> 00:13:11.200 the blueprint here. This automated video factory, 00:13:11.399 --> 00:13:13.779 it's not some future concept. It's deployable, 00:13:13.879 --> 00:13:17.440 like today. By combining Sora 2 access specifically 00:13:17.440 --> 00:13:20.340 through Key AI with an automation tool like NAN, 00:13:20.460 --> 00:13:23.220 you get way more than just the 6x cost savings 00:13:23.220 --> 00:13:25.899 and the watermark removal. You get massive reliable 00:13:25.899 --> 00:13:28.799 scalability. It feels like a fundamental, probably 00:13:28.799 --> 00:13:30.860 permanent change in the economics of creating 00:13:30.860 --> 00:13:34.120 video. That barrier to entry for producing consistent. 00:13:34.730 --> 00:13:38.250 high quality, almost cinematic content, it feels 00:13:38.250 --> 00:13:39.929 like it's been entirely destroyed. It really 00:13:39.929 --> 00:13:42.169 has. And the opportunity right now is just wide 00:13:42.169 --> 00:13:44.289 open. For freelancers, for content creators, 00:13:44.429 --> 00:13:46.769 for businesses small and large, you can build 00:13:46.769 --> 00:13:48.529 really high demand services around this stack. 00:13:48.710 --> 00:13:50.529 Seriously, you should prioritize trying to implement 00:13:50.529 --> 00:13:52.889 this blueprint, maybe even this week. Don't just 00:13:52.889 --> 00:13:54.870 read about it, do it. Okay, final thought then. 00:13:55.190 --> 00:13:58.220 If consistent... Cinematic video content can 00:13:58.220 --> 00:14:00.539 be generated this reliably, this cheaply, like 00:14:00.539 --> 00:14:03.559 1 .5 cents per second. Is it possible that human 00:14:03.559 --> 00:14:05.700 shot single scene video production becomes kind 00:14:05.700 --> 00:14:08.720 of a niche luxury within, say, five years? Maybe 00:14:08.720 --> 00:14:10.419 similar to how the film photography feels today 00:14:10.419 --> 00:14:11.120 or to your own music.