09 January 2025
Corbett Drummey AI in Content Creation & Insights from LTX Studio (262) - E262
In this episode of Digital Marketing Masters, host Matt Rouse welcomes Corbett Drummy from Light Tricks to discuss the intersection of AI and digital marketing. Corbett, co-founder and former CEO of Popular Pays, shares his journey with Light Tricks, known for popular apps like Facetune, and their new AI video product, LTX Studio. The conversation delves into the innovative use of AI in content creation, highlighting how tools like Facetune are used for AI-generated images and localized landing pages to boost marketing efforts.
Corbett and Matt explore the democratization of content creation through AI, emphasizing how small to medium-sized businesses can leverage these tools to enhance their marketing strategies. They discuss the challenges and opportunities AI presents, particularly in maintaining character consistency in video models and the rapid pace of AI advancements.
The discussion also touches on the evolving landscape of social media platforms, the potential for AI-generated content to revolutionize advertising, and the importance of focusing on practical use cases for AI technology. Corbett shares insights into the development of LTX Studio, a tool designed for storyboarding and pre-visualizing video content, and the broader implications of AI in content creation.
As the episode wraps up, Matt and Corbett reflect on the future of AI in marketing and content creation, considering the potential for new platforms and the ongoing evolution of digital media. Corbett invites listeners to explore Light Tricks' suite of products and connect with him on social media for further discussions.
[00:00:02]
Unknown:
Welcome to Digital Marketing Masters. Today, we've got Corbett Drummy from Light ricks with your host, Matt Rouse. Hey, everybody. Welcome back to Digital Marketing Masters. I'm here with Corbett Drummy. How are you, Corbett? Good. Thanks for having me on the show. Hey. Thanks for being on the show. And Corbett is from Lighttrics, and they have a number of products. But, Corbett, why don't you introduce yourself and tell us what you do, and then we can talk about AI? Sure. Absolutely.
[00:00:43] Corbett Drummy:
So I'm Corporate Drummy. The I was the cofounder and CEO of Popular Pays, an influencer and creator marketing platform. And almost 3 years ago now, Lightrix acquired the company. So Lytrix, the parent company, most people know of it through Facetune, its most popular product, but there's a whole suite of products underneath it such as Photo Leap and Video Leap, which are used by a lot of creators for editing photos and videos. But one of the, coolest things to talk about is we released under under electrics a new product called ltx studio, which is an AI video product. And in the last year, I've been spending most of my time on that product on ltx studio. But there's a whole host of things for us to chat through because in in the last few weeks, we've released our own video model powering LTX Studio. We've released some surveys on the industry about AI adoption and the like. So I'm just excited to dig in about all of the above.
[00:01:34] Unknown:
Thanks. And so I've I've already been a subscriber to Facetune. We were using it for some social media editing. Specifically, I actually learned this from a marketer named Perry Belcher, who I'm not sure if you're familiar with. He's a long time kind of direct media marketer,
[00:01:51] Corbett Drummy:
and they were talking about their AI influencers where they were just basically making a AI generated photo and they were using Facetune to kinda tweak the photos so that they're consistent. Yeah. We tried it out. It works great. And, also, you can, like, change makeup and hairstyles. My daughter loves using it. She takes her photo and then changes her hair with it. You know? I I love that use case too of using a, using Facetune and other tools to touch up the AI generated images. Most people, myself included, tend to think of just, you know, you generate content with AI and it's done, but that's a really cool workflow. I hadn't heard much of that. However, we have experimented with using AI generated models and things like localized landing pages, which we have saw we we did see increased conversion on some of our products,
[00:02:34] Unknown:
Have only dipped our toes in that water really. Yeah. I think another cool thing that we did with it was we would change the makeup colors on, like, a stock photo to match the colors of the website. It's really smart. Then you can have a picture in there when it where it matches. Actually, I I stole that from the video game industry. They always have, like, the, you know, like, the interface of the video game will have the character on it, and they'll have, like, you know, purple eye shadow for their purple logo. Mhmm. You know, that kind of thing.
[00:03:03] Corbett Drummy:
Most people think about Facetune as that, you know, b to c consumer app, and that's some those are some really clever business use cases. I hadn't really thought about those. Certainly, one thing's true is that with AI now, marketers can experiment more and the the cost of experimenting has gone down a lot. So you can stand up, you know, so many pieces of content and landing pages and stuff like that just to see what works. You know, content is usually the bottleneck for ads, and AI definitely helps address that.
[00:03:31] Unknown:
I think the biggest thing, you know, for a kind of small to medium sized business audience, which is, you know, most of the people who are listening to this, the there is no cost to like, there's no budget for experimentation. Right? Mhmm. They have a list, and they're like, I got these 27 things that I gotta get done. I only got time to do 15 of them. There's no time to go out and say, okay. I'm gonna start swapping makeup on models for my social media post. Right? Yeah. But I think that's where the kind of digital agency comes in and, you know, the app providers are the ones who need to find the use cases and teach those to people who need them. And, like, I'm not saying everybody should go out and make, you know, make up changes on fake AI models, but, you know, the there's a real problem, I think, with foundational model companies. Like, Google has probably some of the worst demos that I've ever seen for use cases. Like, take this, use our AI to go find a dress you saw in a YouTube video. I'm like, that's not a use case that anyone cares about. Right? Like, what? 1 in a 1000000000 people maybe need that use case? Like, I don't know. I think finding use cases is the most important thing because then your busy person who's a business owner or, you know, they're a real estate agent, they're a trades person, they're whatever it is. Right?
[00:05:00] Corbett Drummy:
They have a task that needs to be done. If you could be like, hey. You know that task that you have that takes half an hour? I can make it 5 minutes. Yeah. That's it. They're sold. They don't give you shit after that. They're like, I'm in. That's a a really important point is when you're building products for small to medium sized businesses or other startups, it really needs to get them to complete and get them to the the finish line. And oftentimes there are tools that help you make something really good, but it's only part of the process. It's not like a completed ad unit for example. And but I do think AI tools will are helping a lot to take these small to mid medium sized businesses or startups and get them to completion and making them good enough to get a final asset. So, you know, an example might be most small businesses, yeah, struggle with content as an example. Right. They don't have anything to post. They might only have statics and now they turn and get, like, get videos out of them. So I think AI is democratizing content creation and making marketers more more well rounded whether you're in house or at an agency. You can be more of, like, a full stack teammate using AI tools. You know, do maybe develop some you know, if you are technical, you could use it to develop some type of website animation that you wouldn't have had the skill to do before. Or even if you're not technical, you people who can't use Adobe and other things can use these AI tools to make content. So they're getting easier and easier for people to take part in.
[00:06:19] Unknown:
Yeah. And there's a real I don't I don't know. I I'm kind of stammering over what I'm trying to say. There I find that there's this real problem right now with companies bolting AI onto their products with no use case whatsoever. Right? But when I look at AI companies, like, you know, your company is essentially an AI company. Right? That AI use case is there before the product was built. So they're like, we can use this to solve these types of problems, not how do I stick AI onto my, you know, CRM so that it looks good on our, you know, to our investors.
[00:07:01] Corbett Drummy:
Right? That is such an important point. So okay. The, you know, last 10 years when when we were running popular pays, we're developing products, we're shipping them to market. That's probably the most important piece of product feedback is that a a lot of teams do solutioneering or they they get they like some technology and they build it, and they're not really focused on that that use case. And to make something effective, you need to start with the use case and work backwards from there and then use the technology that's appropriate. That's probably the most common mistake people make when they're building product. But and it's it's even magnified with AI. Like, this is happening more and more and more. You pointed it out. People are bolting it on. And for example, with AI, there's there's a really huge improvement curve, like an exponential improvement curve. And so you can enter a market with a specific use case knowing that that will expand later, but you have to be really focused on what it's good at today.
One really relevant example that is with our own AI video product, ltx studio. So, you know, imagine you put in a blurb and in, you know, in instead of a single video, you're getting back a full storyboard with shots grouped into scenes and, you know, there's audio dialogue, you know, soundtrack, stuff like that and you can export it as a final video. So it gets you to that endpoint. But still, we knew where the the quality curve was in AI video and we didn't try to pitch it as, like, hey, use this for final pixels in your ads. This is a storyboarding tool. Like, for example, an agency might be pitching something for to a brand and they can use this as a quick story. They can get in, like, 3 minutes storyboard that can successfully communicate their idea and that just improves the creative process that much. Or you can use it to, like, pre visualize, you know, a video effect that you wanna make. So that's where it's it's at today. Yeah. In, like, 6, 9 months, people are already starting to use it for final pixels, but trust me, those models are getting there in the near term. But that's why today we're focused on, like, what is it good at today so that people know that use case that actual the actual use case they have that we can solve it. And I think as entrepreneurs, if you're building product, you you really have to focus on you actually wanna start really small. Most people try to make something for everyone, and you wanna stay ways more focused than you would normally think when you're just starting it up. Yeah.
[00:09:13] Unknown:
I I think honestly that goes for any business that you start is what happens is people if you go too broad, you're trying to make it for everyone and then you end up being not for anyone. Mhmm. Right? We've made that mistake. Yeah. Yeah. I mean, well, it happens in everything. Right? You know? It's it's like the carpenter who says I can build whatever you want. Well, then nobody can find you on the Internet because they're not searching for whatever. It's a great example. Searching for, like, bathroom remodel. Right? Yeah. Well, if you wanna be, I'm the bathroom remodel contractor, then you're easier to find on the Internet. Right? Mhmm. And you remodel 20 people's bathrooms. One of them is gonna be like, I need my kitchen done too. Right? So it all works out in that case. But that I mean, that goes for software too. And It's funny. It ties so much into not not just product, like I said, but marketing, like you said. You know, it's it's all baked in together. That focus will help your product, and it'll also help your marketing.
For sure. So I messed around a bit with LTX. I use it on Hugging Face. Mhmm. For those of you who don't know yeah. Let's let's back up a step for some people who maybe don't know all the jargon that we're using right now. Yeah. Like, when you say final pixel, you mean the output that's gonna be seen by the public. Right? Yes. Fair. So,
[00:10:27] Corbett Drummy:
a lot of people probably the biggest use case of AI right now actually is people experimenting, like enthusiast, stuff like that. There we have seen people using it. There's a lot of good business cases, like people using it for storyboarding and drafting concepts. There are some people using it for what I said, the the final pixels of an ad unit. So we've had you know, we had an an ad created with ltx studio that aired during the Olympics, but that's the small slice of things. Most people were using this for, you know, drafting, prototyping, and moving quickly. But you referenced we have both our AI video platform called ltx studio, which is that workflow where you can get that full storyboard and export a full project. And then you reference ltx b, our video model, which is posted on Hugging Face and other areas where it's actually the foundational video model that powers ltx Studio. So the the video results you're getting in those individual shots, it's powered by l t x v, the foundational model that the team here built.
[00:11:23] Unknown:
Right. I've actually, I found found it pretty good. It seems to be really good with consistency of character kind of thing. Like, if you say, I want a jellyfish, the jellyfish stays a jellyfish. It doesn't turn into, like, half a fish and half a shark or some weird ass thing. Right? And I think that morphing thing is probably one of the most difficult parts of video. You know? It seems to be this like, it's generating frames, you know, for the video, and then it just kind of loses the plot along the way. Mhmm. But video models in general tend to be getting better with I'd say that 1 and a half to 3 second kind of clip staying consistent.
And in, you know, advertising, 3 to 5 seconds is your is your cut. Right?
[00:12:13] Corbett Drummy:
Mhmm. And movies now. I like Yeah. And movies too. Yeah. I'm not technical in a sense of, like, I'm not a designer. The the funny thing about, like, when you're a cofounder of a startup, you're you're not good at anything. You just you do it and then you hire someone who's better than you at everything. So I have no core skill set. But when I was using ltx studio and making my first project, actually, the first thing I did, I we just talked about Isaac Asimov, and my favorite story in the world is just a 3 page story called the last question. And it's kinda about, like, the meaning of the universe and it's wonder it's a wonderful story. It's super short and no one's ever made anything visual for it. So I was, like, oh, I wonder if I can make, like, a 5 minute video of this. And as I started making it, I started watching shows differently. So, like, I saw something on Netflix and it was all these quick, like, the 1 to 3 second cuts. And I was thinking, like, man, I could, like, in 6 months, I could probably make something, like, in in this was this was 6 months ago. And I was thinking, like, where the technology is going. I could probably make something similar in quality in 6 months. And, you know, it took me about 2 hours to really make something I was proud of for, like, a 3 minute clip a 3 minute series of clips with LTX Studio, but you can get a draft in in 3 minutes. That's halfway there. But the way that movies are done, it's taking into account our short tension spans now. And so there's no long shots anymore. And AI is getting better at long shots, but you're right. It's like the the quicker the clip, the less decay there is in the quality. But something we focused on really hard was that character consistency because every model out there at the time we built this was lacking that. And there's lots of progress in that field, but that was something that no one addressed. And so we really try to specialize in that with this. Right. The only other model I think that has good character consistency right now is probably Runway.
[00:13:58] Unknown:
Sora is not good at it a lot of times. Well, I mean, Sora is good at keeping, like, like, the idea of the character through the video, but it keeps changing the way that character looks. Right? Like, an example would be if I have a baby penguin and a mommy penguin and they're on some ice, the baby penguin will, like, turn into a giant penguin. Like, it just doesn't make sense, you know, but it stays a penguin. But you know? So, I mean, yeah, there's weird shit that happens with them right now. And, I mean, all that stuff's gonna go away. Like, g p t if you ever use g p t 3, like, it would say weird shit half the time. Right? And now you got o 3 that can solve math problems I can't even understand.
[00:14:40] Corbett Drummy:
It's it's wild. I mean, I've never worked on something where the pace of change is this rapid, and that's why it's it's almost like you start realizing that content will be very soon a solved problem. And so people are almost, like, not worrying about it. Now I'm just wondering, like, what's next kind of. But when you talk about the consistency thing, yeah, we originally like, l t x studio was built l t x v, sorry, the model was built with the premise of being, like, the fastest, most cost efficient model while still keeping that quality. And so, you know, it's it's the only you can get, like, a 10 second clip with only 1.5 seconds of, like, compute time or, like, run time. Right. So you can generate stuff faster than than it's rendering. So it takes you longer to watch it than it does to generate it. So we knew that these some like, when I use some of the other models, like, you know, Clang or whatever, I had to wait, like, 5, 10 minutes for a clip. And here, you know, you can click a button and render, you know, 80 videos asynchronously in lpx studio. They change the style. You can render it in a few seconds, like, 10, 15 seconds usually.
And so we really wanted to arm people with, like, speed and agility for their creative process. But the quality comparison, you know, considering the hype around Sora and other things, the response to our community was really positive about, like, the quality of our outputs. But you're right. Like, right around the horizon, there's VO, which is, you know, the clips the cherry pick clips of that are amazing and but those get better and better. You know, that's we we should just expect that in our industry that video is gonna be almost perfect in, like, you know, a year, year and a half.
[00:16:08] Unknown:
Right. And I'm working on the second edition of will AI take my job? It's gonna be called will AI take my job 2 with the number 2 at the end.
[00:16:17] Corbett Drummy:
Like I said, it might just be something called AI took my job. That's right. That'll be the final final shot. I'll just stop writing after that. I like
[00:16:26] Unknown:
So one thing actually in the book that I'm talking about that wasn't in the previous version is is I call it the content singularity, and that is where you can generate production quality content faster than it takes to create it and faster than anyone can consume it. So an example of this is, like, Suno. I'm sure you're familiar with the music generator. So it takes Suno if you if you take into account that it's writing the end of the song while you're listening to the beginning, you know, because it streams content. But if you say how long does it take for it to make the entire song, takes about 30 seconds to make a 2 and a half minute, 3 minute song.
Right? So even if you use the API and you're going to, like, chat GBT or something, right, and and you have it write the lyrics and and put it in the format for sudo and send it to sudo and have it generate the song and have it generate 2 of them and then take pick 1 of them and come back. It's already you you're maybe 45 seconds in. But in 45 seconds, you're making close to a production quality song, you know, that's been written and and produced.
[00:17:38] Corbett Drummy:
The and the the content singularly concept is really interesting, especially because, like, with the concept of the broader singularity of kind of runaway AI AI progress, there's a bunch of mini ones that happen in different industries. And arguably, content is the ground 0 of that, like, the epicenter of of that. It's where, you know, like, kind of the first killer app was the Chat GbT with text, and then it quickly moved to images and video and sound. And it's it's hitting other industries like finance, health. Coding is really hit by it and changed by it. But and all these many singularities, my team and I before the acquisition, we we had wondered about this for a while. We just kinda looked at where AI progress was going, and we also knew that, you know, we didn't have, I would say, the the firepower to compete in that world. And that's why one of the reasons I'm so grateful Popular Pay is joined up was acquired by Lightrix was that they do have this world class r and d team. You know, they made their own foundational model.
They have, like, the data and research capabilities here. But with our industry in advertising, which is where we operated, we figured that not only would content approach a singularity, but advertising would. Like, you could conceivably have content being made by these systems and tested with, you know, Facebook ads, for example, and optimized so fast that a human is not even in the loop. And we didn't know exactly when that would happen, but we knew it'd be, you know, in the next 5 years. And it's already possible now, but it's not better than humans in the mix. But probably in the next 5 years, it will be better than humans in the mix in some areas, like some aspects of Facebook Yeah. Ads optimization combined with, you know, dynamic creative made by AI. It's it's pretty wild.
[00:19:15] Unknown:
Yeah. Those it's definitely well, I wouldn't even say it's coming. I'd say it's pretty much here already for most of it except for, you know, maybe the video and music and stuff is not there yet, but the actual automation of the marketing per person, right, like, on a on a one to one model. Mhmm. Like, an example of this is we have a system in a specific industry that customizes the products for them in advance and sends it to them, and all they have to do is say, yes, or I wanna edit this. Mhmm. Like, it already shows them what the product's gonna look like. They don't have to find us, find the product, upload their photo, put in their info, do all the customizing, get a proof. We just send them a proof before they even know that it exists.
[00:20:00] Corbett Drummy:
Right? Swag.
[00:20:01] Unknown:
So and we could do that on a one to one basis, right, for each person. And we can do it at scale. So
[00:20:08] Corbett Drummy:
What's what's interesting on this stuff too is the the early tests might work and then some might really miss. One that missed was Facebook and Instagram tested these AI influencers that were posting kind of bland and really weird inauthentic content. Like, it's something it's, like, really bland content for a human, but it's really inauthentic for an AI. But what's so funny about that is that these could work today, but what you'd wanna do is lean into the glitchiness and the weirdness of AI posting content. So probably the the most arcane and weird example that is, like, truth terminal on Twitter where some AI was hooked up to Twitter and just tweets really crazy stuff. And it's bizarre, but it's more viral and interesting than posting really bland stuff that is safe, you know. And so funny enough, like, Facebook and Instagram, they were testing AI content and it was received terribly. But I still think some AI influencers would work, but you'd actually want them to be kind of off the wall and weirder, and that's what would attract, like. Right. You're never gonna make a better TikTok than, like, the the current ones that exist today in the same style as them. It's just it's so dialed in with those how funny people are, how engaged they are. But you can make something different, and it can catch people's attentions through being just really out there. So, yeah, it's I think people actually need to take more risks and embrace more, like, the big companies, I mean. And for the the small and medium sized businesses, I think just adopting the technology is important, but, yes, it it's probably too hard. I wouldn't bother trying to carve new territory. I think just use these tools to optimize what you're doing today. You know?
[00:21:35] Unknown:
Yeah. I the meta just is one thing after another right now. They're just, like, flushing their platform down the toilet right now. I don't know what the fuck they're thinking. Like, honestly, like, I posted this this photo today, and it there's something to the effect of Facebook founder dies at age 36 of syphilis after saying misinformation will not be from our platform. You know? Like like, we're not gonna fat check it, but, by the way, he's dead. You know? Like, it and it's obviously untrue. Yeah. But, yeah, like, the whole thing is, like, okay. So they went, okay. We're not gonna allow virtual influencers, and they were like, okay. Yeah. We'll allow it. And then they're like, anybody who has more than one profile, we're gonna delete it. And if it's not a real person for you, we're gonna delete it. And then they're like, now we're gonna put our own ones of what we said where you're not allowed to do. We're gonna do it.
And then, of course, they were terrible because, like, no one wants to talk to AI Snoop Dogg. They actually just wanna talk to real Snoop Dogg. You know? Like and then they did the one where they were AI influencers that weren't real people, but they're pretending to be, like, average people. Yeah.
[00:22:45] Corbett Drummy:
That's the one I was referencing.
[00:22:46] Unknown:
Nobody wants to talk to an average
[00:22:49] Corbett Drummy:
fake average person. Like, most people don't wanna talk to an average person to begin with. How is there no one. Was about saying. I think that you're in the market for a start up to make a bunch of, like, kinda like, let let AI run loose, you know, not not like a Right. Not in that sense of I I mean, just you leveraging the the models that are out there today and, you know, already tested and saved. But letting them be more creative and do things differently than humans would. But using them to try to mimic the most average human is exactly what a large company like Facebook would do and is just, of course, gonna not work out.
[00:23:19] Unknown:
Yeah. It's just a complete misstep of, you know, who are our users? What do they want from our platform? I I don't know. I think it's obvious. You can take a small group of people pretty much anywhere, put them in a room with a couple of beers or a cup of coffee or whatever and say, what do you use Facebook for? And they're like, I wanna see events. I use Facebook Marketplace. I wanna know when Jim's having his baby. I wanna know when I gotta get to the birthday party. And, you know, like, they're not like, I really wish I could talk to a fake version of Bob. Like I know. I know.
[00:23:55] Corbett Drummy:
It's I think that, it shows why, like, acquiring startups when you're that big, like, a Facebook size is important because it brings, like, fresh blood and ideas into the company, but that's m and a has been a little bit cool off for various reasons, regulatory, economic, etcetera. But when you're just such a big company, it's really hard to take the risk that are required for, like, these new product things. However, you can take advantage of some things like the sheer scale of Facebook. Let it to be successful with things like, you know, they're the 1st people to really attack, like, the Craigslist with Facebook Marketplace, which is a great product. You know, people use it all the time. But, one thing They also they also said, you know, oh, there's been, like, 400,000,000
[00:24:33] Unknown:
AI uses on Facebook already, but they feed it to you without you asking. That's not that's not a use of AI. Right? That's that's the opposite of having people want AI. It's like it's I don't know. Like, it it's just the the opposite of getting people to use something is forcing them to have to use it. Yeah. Right? And that's what you totally do not wanna do. One thing that they are good at though and the these, like, mega companies, you know, TikTok, Facebook, you know, Meta as a whole, etcetera, Twitter and x.
[00:25:05] Corbett Drummy:
The algorithmic feeds will help with what you pointed out with the content singularity because we're already creating more content than we could consume, and that's only gonna be exacerbated. And that would be too impossible with following based feeds. But with algorithmic feeds, yeah, there'll be way more AI slop and bad content than ever before, but there'll be, you know, some content that's 10 x better just because the sheer content you'll have, the outliers that are real winners. And those algorithms are great at surfacing that. So I think everything shifted over. You know, TikTok really pushed it there. Others have tried it before to some success, but that really, everyone started shifting over after, you know, TikTok in the COVID era. Now it's it's complete. It's, like, all algorithmic everywhere. But that's really the only way you can handle the influx of content we'll see from AI. But, yeah, I think Facebook's, like, putting a thumb on the scale with, like, hey. They're force feeding these fake things to you, and that's they're kind of ruining their own algorithm there. I think things are gonna switch back.
[00:25:59] Unknown:
I think the pendulum is gonna swing back from heavy algorithmic content favored by the company Mhmm. To favored by the user. And, like, TikTok's algorithm originally was really good at saying, oh, okay. Well, you seem to like these types of things. You know? You seem to like gaming stuff, so we're gonna show you more gaming stuff. Mhmm. Right? But at this point, I mean, I go look at these apps, and there is very little that I am interested in that it shows me. Yeah. And what's happened is it's it's kind of swayed towards demographics, like really large generalized demographics.
And, you know, I get the the worst stuff being a 50 something year old white man. I get, like, do I wanna date girls from overseas, or do you want to, like here's here's 50 different Facebook groups that basically are just pictures of boobs. Like, this is just stuff that does not interest me. Not that I'm not interested in that, but it's not what I want on Facebook. Right? No. You know? So, like, it's just what I think is gonna happen is you're gonna have to have a way for people to start to make selections, like obvious selections, like do you want more of this? Yes or no? Yes. Right? And people will identify what they want, and then the algorithm can feed their preference that they have chosen, not a preference that we have given them a 100 different choices of things they don't want, and they have we pick the 10 that they watched the most,
[00:27:36] Corbett Drummy:
you know, which are still not things that they like. I I totally agree. The I think there there's some some that I've just stopped using recently because of the algorithms are so frustrating to me. Like, I've just stopped using Instagram, like, the Explorer and real reels are really frustrating to me. And sometimes it's funny how transparent it is where, like, if you, you know, like, one soccer video, it's like, man, you really love soccer. Here's a ton of soccer. And I've seen people gain their feeds. Like, the momentum is so sensitive. I I've seen people gain their algorithmic feeds where they just will, like, only, you know, images of people cooking with eggs and all of a sudden their entire feed is just eggs and stuff. And it lasts the short term, but the the frustrating thing is, like, when you might get, like, a funny meme like, I usually pop on. I just get, like, random general demographic memes and stuff. But then, it's not what I want long term, and I want to have that conscious choice of, like, steering it. Like, I wanna discover stuff in certain areas, and there's no way for me to push that. I don't get served much content in that area in general. And so I can't really lean into it. I don't get served much of it, and there's no, like, stated preferences. So I've And you can't search on Meta anymore because the AI system just tells you, like, a definition of what you're trying to search for instead of actually searching. And with your note about, like, Facebook itself, like, I'm sure that's just complete garbage over there. I haven't really logged it. Like, I probably pop it open once every 2 months and yeah. So, fortunately, I do usually see on my feed, like, oh, someone got engaged or had a had a baby, etcetera. And so a meaningful life update, but I I just have lost all utility from their products. So, like, I've I really have only used it once in the last few months for, like, maybe Google our Facebook Marketplace and once for, you know, just I don't know. I I just have kind of
[00:29:10] Unknown:
completely left. I would say for the first time in 17 years, I've bumped Facebook off the front page of my phone to a Yep. Secondary page because I don't use it anymore. I am liking Blue Sky a lot. Yeah. What's your experience been like on that? I've never I haven't did this ever since. Really, really good, actually. It has a good way for you to kinda start out figuring out who to follow. Yeah. And it is diametrically opposed to when you try to make a new account on x, which I just had to do for someone actually 2 days ago. And they're just like, do you wanna follow technology, sports, or Elon Musk? Like, these are, like, these choices at the beginning. Like, it's just it's completely made for they're like, what do you want us to show you that we're gonna decide on, not what do you want? Yeah. They so I I still probably use Twitter access my most
[00:30:01] Corbett Drummy:
used thing, but it's because, like, I've intentionally like, I get all my AI news from that, and I've intentionally followed people. Right. And every once in a while, I'll click over to the Explore feed, but the the the the amount they push Elon is so freaking frustrating. It's like I had enough of it so quickly and banned him, but it it still drives so much of the discourse. But I don't know. That that website is crazy, but I still need some outlet for finding some kind of new source, you know, like Right. I'm Well, Blue Sky Blue Sky allows links.
[00:30:31] Unknown:
Oh, that that's helpful. And they don't downgrade links. So Yeah. Somebody posts a link of, hey. Here's an you know, here's some deep journalistic
[00:30:41] Corbett Drummy:
story that we've done investigative journalism about this topic and there's a link to it, then people actually see it. Alright. Whereas on every other platform, they do not. You've gotten me down the the the marketing funnel there. I'm I'm now full like, I will trial it now. Like, I've heard it from enough people, you know. It's like influencer marketing. Like, you know, I've heard it from fam like, my uncle or friends, and now I will, cross the chasm and and try it out. Like, I tried threads, and it just didn't didn't really fit. So You guys can all follow me on there. I will. I will. I'll give it a I'll give it a try.
[00:31:13] Unknown:
My, my handle is author mattraus dothookdm.com. Cool. Alright. Well, you can actually use your own domain as well. So you could you could verify your account through your own domain. Yeah. It's just it's like starting up new networks is hard. It takes that activation effort.
[00:31:29] Corbett Drummy:
And but I do think also one I'd love to hear your predictions actually in your book, like, some that are coming up. But one is that usually with new ways of creating content, there's usually a new network. And so I I do fully expect that with the progress of AI, like, we'll see a new, like, family guy or Rick and Morty style thing that is really beloved and takes hold. And I don't know if that's gonna happen on on YouTube or, like, a totally new network. You know, it's I see lots of cool things created in our community. Yeah. And but, like, I I wonder what's gonna happen there, but there might it might be something in between a Netflix and a TikTok, like, similar way to stream like, get good content. But Hi. I don't know. But I I'm fully expecting with just, like, the revolution in in AI to breed a new way of consuming content. But I don't know if I think blue sky might be a slow burn that grows over time, but we'll probably see something rocket up from from 0 in the next, you know, year or 2 as it gets to be so easy to create and produce and and generate content that rivals, you know, kinda low budget, you know, animated series and stuff like that. Like, you can already create a family guy with with this now, but it's gonna be democratizing so easy that I think it'll we'll see new types of content being made soon.
[00:32:39] Unknown:
I think the first mainstream, like, single or small group project using AI is gonna be out this year. Probably soon. Like, within months, maybe weeks. Probably before my book comes out. It's just it's too it's too easy to do now. Right? Yeah. And with the editing tools and things that are out now, also with the ability to do stuff like like shoot live and convert to cartoon,
[00:33:07] Corbett Drummy:
you know, with, like, runway or something. Right? It's it's too easy to make it. So that it's a writing problem now. Yeah. We like, the thing that really can that came into me is when when I use style reference on our ltx studio product where you can make something and then with a single image, change the whole style of that project to be that image. You can, as a creator, get something on, like, I don't know, unsplash that or something that's, you know, you have the rights to or you can make something in your own style and then makes convert the whole project to that style and it pushes it out of that bland default AI, you know, kind of feel. But that way, you can make something distinctive and cool. And it doesn't need like, there's tons of hits that aren't fully perfect in their animation. Again, like, from Rick and Morty to anime or whatever, you can do that today. But it's it's Right. It's just hit like, I think the overhang of what tech can do versus what consumers have adopted is still pretty great and will only increase. But I think as that technology is be being adopted, probably once you can easily do it just from your phone, I really think we're gonna see an explosion of of content creation. It's gonna shift from, like, the, you know, the the Netflix is spending billions on content to these up and coming creators making full series. You know? Yeah. Yeah.
[00:34:18] Unknown:
And I don't think the the production cost is gonna be low enough to make something high enough quality that you could put it on Netflix or one of the other networks. And I think it's probably gonna come to the other networks first. It's gonna come to, like, Tubi or, you know, what was freebie kind of prime or whatever it is now, whatever they call the freebie version. There's so many names. I know. Roku and stuff like that. Right? It's where you can basically as long as you can pass the quality guideline, you can get on the network. Right? So as long as the people who watch it say, okay. This is good enough for our network that it's on. And then it's gonna be someone is gonna make something that is, you know, some one single creator or a couple a small team are gonna have some funny people mixed with some technically savvy people, and they're gonna put out something that's just gonna blow the doors off everything in Hollywood, and they're gonna shit their pants. I agree. Right? The great irony is that Creepy, which is one of the biggest flops of all time,
[00:35:16] Corbett Drummy:
is probably right, just wrong timing. Like, I honestly think there's a market for that stuff. Initially, TikTok ate it all up during COVID, and it's just wrong timing. You you didn't needed to go big blockbuster, like, HBO, Netflix style funded stuff, high quality, and, like, a little bit longer format or short form driven by creators from the ground up. But I really think, like you said, there's gonna be a single creator or small group make a real hit show or series or just, like, you know, comedy a couple short comedy sketches or whatever. And but I think that will end end up birthing a new platform followed by intense competition from the big ones.
[00:35:52] Unknown:
Yeah. I think one of the things that's probably gonna come out is some kind of streaming service similar to YouTube. Yeah. But it will probably be some sort of privately held group of creators who make a show or multiple shows, kind of like the Adult Swim idea Mhmm. Where it's it's all under the same umbrella. But the reason that I think it's gonna be done separately is because of the, like, the the supreme court turning over the FCC ruling in the United States about, you know, the Internet being I forget what the wording for it is. The net neutrality thing? Yeah. The net neutrality. So it will be easy for large cable networks to say, you can on your Internet package, these are the only streaming services you can use. Mhmm. Because they'll just block traffic to the other ones. Yeah.
[00:36:43] Corbett Drummy:
You can use. Mhmm. Because they'll just block traffic to the other ones. Yeah. I'm not really exposed much to the like, we're gonna have some downstream effects, like, you know, political and regulatory effects in the tech world. I'm not I'm not an expert in that area of, like, what we can expect. But I do think it's at least more open for innovation in the US market. I think Europe will lag behind a little bit with you know, they're already there's already a lot of AI products that are not rolling out in Europe because of certain regulations. And so Right. I think they're gonna lag behind a bit. It's a valiant effort to do certain things. And, like, because of things like g GDPR, it's changed how everyone approaches data. But it's it might be a little bit too early to have regulated yet.
You know, it's like regulating too early. You get stuck with the
[00:37:26] Unknown:
USB versus USB c or, like, inferior things for a long time. And but, yeah, I I I don't know exactly what's gonna happen in that neutrality and the downstream effects of all that. But Well, that's why I think things are gonna come off of kind of major networks and and be done outside of the current kind of large tech walled garden areas because they're not gonna be regulated. Right? It's the same idea as TikTok's new app, Lemon 8, I think it's called. Just Lemon with the number 8. Yeah. I haven't used it. It's basically just TikTok again. Like, it has some other features to it. Yeah. I had just downloaded it today. But so if they ban TikTok, they everybody from TikTok's already going over the other app anyway. Right? So they'll just, like, what are they gonna turn around and ban that app too? Like, I'm everyone that I was thinking of
[00:38:09] Corbett Drummy:
it. Like, they're gonna have that one day of ban, you know, the 9th January 19th coming up. I'm curious to see what'll happen there. But I would say in general, although TikTok's going through some of those, like, wavering a little bit, I think eventually they'll they'll be fine. But Yeah. I think they'll be fine too. But they I think TikTok and YouTube are best positioned to capture the influx of AI content. Right now old now. Right? Like, how long has TikTok been around? 8 years or so? It's like it'll be them versus a totally new network, but, they've got some advantages. But the the thing is But if you're 12,
[00:38:38] Unknown:
TikTok is what your parents use.
[00:38:40] Corbett Drummy:
Right? So you don't wanna be on TikTok. You're gonna be on whatever the next thing is. Yeah. Yeah. That'll be the next thing. It usually starts with people that have the tech in their hands and a lot of time, like high school and stuff like that. But I do think that the most interesting content right now is being shared actually on, like, Discord and private channels and stuff like that. Yeah. Like, the coolest AI content I've seen is is actually is kind of siloed. And so once that escapes and is being shared on feeds that could be, like, reshared, forked as projects, and contributed to, that that I think that's why I've been saying I think there's a market for that kind of new type of content, and I think it'll start to shift once we're already getting there. Like, I've seen some really cool stuff internally. But once you can just kind of be chat it's we talked about, like, GPT Canvas, and there's some stuff we're exploring. But once you can kind of just chat with these things as your creative assistant and it's like an agent, you don't really need the same interfaces as before, and they're they're going away they're going by the wayside. Like, I am not a coder, but but I can type in English and let gbt canvas code next to me and make a web app. And, like, you've mentioned you use it for writing when we were talking earlier.
But now I think it'll be the same thing about with video. Like, agents will do a lot of the creative production. Like, hey. I want this scene to look like this, this scene to look like that. And once that's kind of in our hands, like, mobile, I think that's when it'll explode, and that's when we'll I think we'll see this the surge of the new networks taking over.
[00:39:55] Unknown:
I think also, something that's definitely gonna be an up and comer is, like, an AI agent kind of, I guess, technically an agent that writes and creates and produces and exports its own show, right, where somebody builds the framework around it Yeah. And then it just makes shows. And once you once you have a system that can do it, right, then you make a better one and a better one and a better one. And soon somebody's watching this show and says, wow. This is actually a pretty good show. And then it takes off. Right? So I would say yes, but I think the intermediate step is gonna be just users prompting the agent to make edits. So I think that'll be like a as well. Yeah. That'll that'll last probably for a while. But we're gonna get there. Like, if anyone here has read, you know, Infinite Jest and there's, like, the the content that gets so addicting that it's hard to put down, and if you take away from someone, they get completely erased and just catatonic.
[00:40:48] Corbett Drummy:
I think there's that singularity of content where eventually, like you said, a AI agents will do do it without humans in the loop, and the content will just get recursively better and better and super addicting. And it might be that kind of a Wall E world where we're all in our VRs just in a kinda kinda bait, you know, on our floating hover beds. We like the glasses from the old RoboCop movie, you know. It's it's like a drug that is Who knows? Maybe we're secretly people are passing them around Yeah. Selling selling content. They'll be, like, in the back alley. They're like, you want some AI? Maybe that's our experience now is yeah, Matt. We're we're we're, we're actually in that right now. So who knows? It's right. We're on the way. It is the most interesting times to be both in I think the epicenter of all this progress is in the content world, so it's crazy to be in this space. But it's it's also while it's happening during our lifetimes, there's gonna be a lot of change and and problems, but there's just hopefully gonna be a lot of abundance and positive things as well.
[00:41:39] Unknown:
Yeah. Well, we gotta hope for the best because, I mean, it's coming whether we like it or not. So it's nice to be getting in there. Hope for the best, prepare for the worst. Right. And, honestly, I I have a pretty pretty cheery outlook on AI over the long term. Me too. So and, I mean, we got a lot of problems in the world. Right? We got a lot of stuff needs solving, and, you know, some of these problems are massively complicated, and and computers are good at that. So,
[00:42:06] Corbett Drummy:
you know, we'll see how everything goes. Corbit, we can talk about this all day, but we should probably wrap it up here. We've been already talking for, like, 40 minutes now. If people wanna reach out to you, what's the best place for them to find you? So I'm at corporate drummy on all socials. So you can just look at me on everything from, you know, Twitter x to LinkedIn. It's just at corporate drummy, Instagram, etcetera. And if you wanna check out the products we talked about, go to light tricks.com. You can see our suite of products there. If you're a creator, definitely try out LTX Studio for making these projects we mentioned. But, yeah, we'd love to chat with anyone if this stuff sparks interest. And, yeah, Matt, thanks for having me on it. Conversation's always fascinating, and we could keep going, but I think, yeah, no one wants to hear us continue to ramble. I think we packed it all in here. So thank you for having me. Let there be light. Yeah. I I like what you did there. You can read that Isaac Asimov story if you didn't catch the reference, the last question. It's a great one.
[00:42:58] Unknown:
Thank you for listening to the digital marketing masters podcast with your host, Matt Rausz. If you enjoyed the show, please leave us an honest review on your favorite podcast app. Be sure to check out Matt's book, will AI take my job, and have a wonderful day. Welcome to digital marketing masters.
[00:43:24] Unknown:
Today, we've got Corbett Drummy from Light Tricks with your host, Matt Rouse. With your host, with your host, Matt Rouse.
Welcome to Digital Marketing Masters. Today, we've got Corbett Drummy from Light ricks with your host, Matt Rouse. Hey, everybody. Welcome back to Digital Marketing Masters. I'm here with Corbett Drummy. How are you, Corbett? Good. Thanks for having me on the show. Hey. Thanks for being on the show. And Corbett is from Lighttrics, and they have a number of products. But, Corbett, why don't you introduce yourself and tell us what you do, and then we can talk about AI? Sure. Absolutely.
[00:00:43] Corbett Drummy:
So I'm Corporate Drummy. The I was the cofounder and CEO of Popular Pays, an influencer and creator marketing platform. And almost 3 years ago now, Lightrix acquired the company. So Lytrix, the parent company, most people know of it through Facetune, its most popular product, but there's a whole suite of products underneath it such as Photo Leap and Video Leap, which are used by a lot of creators for editing photos and videos. But one of the, coolest things to talk about is we released under under electrics a new product called ltx studio, which is an AI video product. And in the last year, I've been spending most of my time on that product on ltx studio. But there's a whole host of things for us to chat through because in in the last few weeks, we've released our own video model powering LTX Studio. We've released some surveys on the industry about AI adoption and the like. So I'm just excited to dig in about all of the above.
[00:01:34] Unknown:
Thanks. And so I've I've already been a subscriber to Facetune. We were using it for some social media editing. Specifically, I actually learned this from a marketer named Perry Belcher, who I'm not sure if you're familiar with. He's a long time kind of direct media marketer,
[00:01:51] Corbett Drummy:
and they were talking about their AI influencers where they were just basically making a AI generated photo and they were using Facetune to kinda tweak the photos so that they're consistent. Yeah. We tried it out. It works great. And, also, you can, like, change makeup and hairstyles. My daughter loves using it. She takes her photo and then changes her hair with it. You know? I I love that use case too of using a, using Facetune and other tools to touch up the AI generated images. Most people, myself included, tend to think of just, you know, you generate content with AI and it's done, but that's a really cool workflow. I hadn't heard much of that. However, we have experimented with using AI generated models and things like localized landing pages, which we have saw we we did see increased conversion on some of our products,
[00:02:34] Unknown:
Have only dipped our toes in that water really. Yeah. I think another cool thing that we did with it was we would change the makeup colors on, like, a stock photo to match the colors of the website. It's really smart. Then you can have a picture in there when it where it matches. Actually, I I stole that from the video game industry. They always have, like, the, you know, like, the interface of the video game will have the character on it, and they'll have, like, you know, purple eye shadow for their purple logo. Mhmm. You know, that kind of thing.
[00:03:03] Corbett Drummy:
Most people think about Facetune as that, you know, b to c consumer app, and that's some those are some really clever business use cases. I hadn't really thought about those. Certainly, one thing's true is that with AI now, marketers can experiment more and the the cost of experimenting has gone down a lot. So you can stand up, you know, so many pieces of content and landing pages and stuff like that just to see what works. You know, content is usually the bottleneck for ads, and AI definitely helps address that.
[00:03:31] Unknown:
I think the biggest thing, you know, for a kind of small to medium sized business audience, which is, you know, most of the people who are listening to this, the there is no cost to like, there's no budget for experimentation. Right? Mhmm. They have a list, and they're like, I got these 27 things that I gotta get done. I only got time to do 15 of them. There's no time to go out and say, okay. I'm gonna start swapping makeup on models for my social media post. Right? Yeah. But I think that's where the kind of digital agency comes in and, you know, the app providers are the ones who need to find the use cases and teach those to people who need them. And, like, I'm not saying everybody should go out and make, you know, make up changes on fake AI models, but, you know, the there's a real problem, I think, with foundational model companies. Like, Google has probably some of the worst demos that I've ever seen for use cases. Like, take this, use our AI to go find a dress you saw in a YouTube video. I'm like, that's not a use case that anyone cares about. Right? Like, what? 1 in a 1000000000 people maybe need that use case? Like, I don't know. I think finding use cases is the most important thing because then your busy person who's a business owner or, you know, they're a real estate agent, they're a trades person, they're whatever it is. Right?
[00:05:00] Corbett Drummy:
They have a task that needs to be done. If you could be like, hey. You know that task that you have that takes half an hour? I can make it 5 minutes. Yeah. That's it. They're sold. They don't give you shit after that. They're like, I'm in. That's a a really important point is when you're building products for small to medium sized businesses or other startups, it really needs to get them to complete and get them to the the finish line. And oftentimes there are tools that help you make something really good, but it's only part of the process. It's not like a completed ad unit for example. And but I do think AI tools will are helping a lot to take these small to mid medium sized businesses or startups and get them to completion and making them good enough to get a final asset. So, you know, an example might be most small businesses, yeah, struggle with content as an example. Right. They don't have anything to post. They might only have statics and now they turn and get, like, get videos out of them. So I think AI is democratizing content creation and making marketers more more well rounded whether you're in house or at an agency. You can be more of, like, a full stack teammate using AI tools. You know, do maybe develop some you know, if you are technical, you could use it to develop some type of website animation that you wouldn't have had the skill to do before. Or even if you're not technical, you people who can't use Adobe and other things can use these AI tools to make content. So they're getting easier and easier for people to take part in.
[00:06:19] Unknown:
Yeah. And there's a real I don't I don't know. I I'm kind of stammering over what I'm trying to say. There I find that there's this real problem right now with companies bolting AI onto their products with no use case whatsoever. Right? But when I look at AI companies, like, you know, your company is essentially an AI company. Right? That AI use case is there before the product was built. So they're like, we can use this to solve these types of problems, not how do I stick AI onto my, you know, CRM so that it looks good on our, you know, to our investors.
[00:07:01] Corbett Drummy:
Right? That is such an important point. So okay. The, you know, last 10 years when when we were running popular pays, we're developing products, we're shipping them to market. That's probably the most important piece of product feedback is that a a lot of teams do solutioneering or they they get they like some technology and they build it, and they're not really focused on that that use case. And to make something effective, you need to start with the use case and work backwards from there and then use the technology that's appropriate. That's probably the most common mistake people make when they're building product. But and it's it's even magnified with AI. Like, this is happening more and more and more. You pointed it out. People are bolting it on. And for example, with AI, there's there's a really huge improvement curve, like an exponential improvement curve. And so you can enter a market with a specific use case knowing that that will expand later, but you have to be really focused on what it's good at today.
One really relevant example that is with our own AI video product, ltx studio. So, you know, imagine you put in a blurb and in, you know, in instead of a single video, you're getting back a full storyboard with shots grouped into scenes and, you know, there's audio dialogue, you know, soundtrack, stuff like that and you can export it as a final video. So it gets you to that endpoint. But still, we knew where the the quality curve was in AI video and we didn't try to pitch it as, like, hey, use this for final pixels in your ads. This is a storyboarding tool. Like, for example, an agency might be pitching something for to a brand and they can use this as a quick story. They can get in, like, 3 minutes storyboard that can successfully communicate their idea and that just improves the creative process that much. Or you can use it to, like, pre visualize, you know, a video effect that you wanna make. So that's where it's it's at today. Yeah. In, like, 6, 9 months, people are already starting to use it for final pixels, but trust me, those models are getting there in the near term. But that's why today we're focused on, like, what is it good at today so that people know that use case that actual the actual use case they have that we can solve it. And I think as entrepreneurs, if you're building product, you you really have to focus on you actually wanna start really small. Most people try to make something for everyone, and you wanna stay ways more focused than you would normally think when you're just starting it up. Yeah.
[00:09:13] Unknown:
I I think honestly that goes for any business that you start is what happens is people if you go too broad, you're trying to make it for everyone and then you end up being not for anyone. Mhmm. Right? We've made that mistake. Yeah. Yeah. I mean, well, it happens in everything. Right? You know? It's it's like the carpenter who says I can build whatever you want. Well, then nobody can find you on the Internet because they're not searching for whatever. It's a great example. Searching for, like, bathroom remodel. Right? Yeah. Well, if you wanna be, I'm the bathroom remodel contractor, then you're easier to find on the Internet. Right? Mhmm. And you remodel 20 people's bathrooms. One of them is gonna be like, I need my kitchen done too. Right? So it all works out in that case. But that I mean, that goes for software too. And It's funny. It ties so much into not not just product, like I said, but marketing, like you said. You know, it's it's all baked in together. That focus will help your product, and it'll also help your marketing.
For sure. So I messed around a bit with LTX. I use it on Hugging Face. Mhmm. For those of you who don't know yeah. Let's let's back up a step for some people who maybe don't know all the jargon that we're using right now. Yeah. Like, when you say final pixel, you mean the output that's gonna be seen by the public. Right? Yes. Fair. So,
[00:10:27] Corbett Drummy:
a lot of people probably the biggest use case of AI right now actually is people experimenting, like enthusiast, stuff like that. There we have seen people using it. There's a lot of good business cases, like people using it for storyboarding and drafting concepts. There are some people using it for what I said, the the final pixels of an ad unit. So we've had you know, we had an an ad created with ltx studio that aired during the Olympics, but that's the small slice of things. Most people were using this for, you know, drafting, prototyping, and moving quickly. But you referenced we have both our AI video platform called ltx studio, which is that workflow where you can get that full storyboard and export a full project. And then you reference ltx b, our video model, which is posted on Hugging Face and other areas where it's actually the foundational video model that powers ltx Studio. So the the video results you're getting in those individual shots, it's powered by l t x v, the foundational model that the team here built.
[00:11:23] Unknown:
Right. I've actually, I found found it pretty good. It seems to be really good with consistency of character kind of thing. Like, if you say, I want a jellyfish, the jellyfish stays a jellyfish. It doesn't turn into, like, half a fish and half a shark or some weird ass thing. Right? And I think that morphing thing is probably one of the most difficult parts of video. You know? It seems to be this like, it's generating frames, you know, for the video, and then it just kind of loses the plot along the way. Mhmm. But video models in general tend to be getting better with I'd say that 1 and a half to 3 second kind of clip staying consistent.
And in, you know, advertising, 3 to 5 seconds is your is your cut. Right?
[00:12:13] Corbett Drummy:
Mhmm. And movies now. I like Yeah. And movies too. Yeah. I'm not technical in a sense of, like, I'm not a designer. The the funny thing about, like, when you're a cofounder of a startup, you're you're not good at anything. You just you do it and then you hire someone who's better than you at everything. So I have no core skill set. But when I was using ltx studio and making my first project, actually, the first thing I did, I we just talked about Isaac Asimov, and my favorite story in the world is just a 3 page story called the last question. And it's kinda about, like, the meaning of the universe and it's wonder it's a wonderful story. It's super short and no one's ever made anything visual for it. So I was, like, oh, I wonder if I can make, like, a 5 minute video of this. And as I started making it, I started watching shows differently. So, like, I saw something on Netflix and it was all these quick, like, the 1 to 3 second cuts. And I was thinking, like, man, I could, like, in 6 months, I could probably make something, like, in in this was this was 6 months ago. And I was thinking, like, where the technology is going. I could probably make something similar in quality in 6 months. And, you know, it took me about 2 hours to really make something I was proud of for, like, a 3 minute clip a 3 minute series of clips with LTX Studio, but you can get a draft in in 3 minutes. That's halfway there. But the way that movies are done, it's taking into account our short tension spans now. And so there's no long shots anymore. And AI is getting better at long shots, but you're right. It's like the the quicker the clip, the less decay there is in the quality. But something we focused on really hard was that character consistency because every model out there at the time we built this was lacking that. And there's lots of progress in that field, but that was something that no one addressed. And so we really try to specialize in that with this. Right. The only other model I think that has good character consistency right now is probably Runway.
[00:13:58] Unknown:
Sora is not good at it a lot of times. Well, I mean, Sora is good at keeping, like, like, the idea of the character through the video, but it keeps changing the way that character looks. Right? Like, an example would be if I have a baby penguin and a mommy penguin and they're on some ice, the baby penguin will, like, turn into a giant penguin. Like, it just doesn't make sense, you know, but it stays a penguin. But you know? So, I mean, yeah, there's weird shit that happens with them right now. And, I mean, all that stuff's gonna go away. Like, g p t if you ever use g p t 3, like, it would say weird shit half the time. Right? And now you got o 3 that can solve math problems I can't even understand.
[00:14:40] Corbett Drummy:
It's it's wild. I mean, I've never worked on something where the pace of change is this rapid, and that's why it's it's almost like you start realizing that content will be very soon a solved problem. And so people are almost, like, not worrying about it. Now I'm just wondering, like, what's next kind of. But when you talk about the consistency thing, yeah, we originally like, l t x studio was built l t x v, sorry, the model was built with the premise of being, like, the fastest, most cost efficient model while still keeping that quality. And so, you know, it's it's the only you can get, like, a 10 second clip with only 1.5 seconds of, like, compute time or, like, run time. Right. So you can generate stuff faster than than it's rendering. So it takes you longer to watch it than it does to generate it. So we knew that these some like, when I use some of the other models, like, you know, Clang or whatever, I had to wait, like, 5, 10 minutes for a clip. And here, you know, you can click a button and render, you know, 80 videos asynchronously in lpx studio. They change the style. You can render it in a few seconds, like, 10, 15 seconds usually.
And so we really wanted to arm people with, like, speed and agility for their creative process. But the quality comparison, you know, considering the hype around Sora and other things, the response to our community was really positive about, like, the quality of our outputs. But you're right. Like, right around the horizon, there's VO, which is, you know, the clips the cherry pick clips of that are amazing and but those get better and better. You know, that's we we should just expect that in our industry that video is gonna be almost perfect in, like, you know, a year, year and a half.
[00:16:08] Unknown:
Right. And I'm working on the second edition of will AI take my job? It's gonna be called will AI take my job 2 with the number 2 at the end.
[00:16:17] Corbett Drummy:
Like I said, it might just be something called AI took my job. That's right. That'll be the final final shot. I'll just stop writing after that. I like
[00:16:26] Unknown:
So one thing actually in the book that I'm talking about that wasn't in the previous version is is I call it the content singularity, and that is where you can generate production quality content faster than it takes to create it and faster than anyone can consume it. So an example of this is, like, Suno. I'm sure you're familiar with the music generator. So it takes Suno if you if you take into account that it's writing the end of the song while you're listening to the beginning, you know, because it streams content. But if you say how long does it take for it to make the entire song, takes about 30 seconds to make a 2 and a half minute, 3 minute song.
Right? So even if you use the API and you're going to, like, chat GBT or something, right, and and you have it write the lyrics and and put it in the format for sudo and send it to sudo and have it generate the song and have it generate 2 of them and then take pick 1 of them and come back. It's already you you're maybe 45 seconds in. But in 45 seconds, you're making close to a production quality song, you know, that's been written and and produced.
[00:17:38] Corbett Drummy:
The and the the content singularly concept is really interesting, especially because, like, with the concept of the broader singularity of kind of runaway AI AI progress, there's a bunch of mini ones that happen in different industries. And arguably, content is the ground 0 of that, like, the epicenter of of that. It's where, you know, like, kind of the first killer app was the Chat GbT with text, and then it quickly moved to images and video and sound. And it's it's hitting other industries like finance, health. Coding is really hit by it and changed by it. But and all these many singularities, my team and I before the acquisition, we we had wondered about this for a while. We just kinda looked at where AI progress was going, and we also knew that, you know, we didn't have, I would say, the the firepower to compete in that world. And that's why one of the reasons I'm so grateful Popular Pay is joined up was acquired by Lightrix was that they do have this world class r and d team. You know, they made their own foundational model.
They have, like, the data and research capabilities here. But with our industry in advertising, which is where we operated, we figured that not only would content approach a singularity, but advertising would. Like, you could conceivably have content being made by these systems and tested with, you know, Facebook ads, for example, and optimized so fast that a human is not even in the loop. And we didn't know exactly when that would happen, but we knew it'd be, you know, in the next 5 years. And it's already possible now, but it's not better than humans in the mix. But probably in the next 5 years, it will be better than humans in the mix in some areas, like some aspects of Facebook Yeah. Ads optimization combined with, you know, dynamic creative made by AI. It's it's pretty wild.
[00:19:15] Unknown:
Yeah. Those it's definitely well, I wouldn't even say it's coming. I'd say it's pretty much here already for most of it except for, you know, maybe the video and music and stuff is not there yet, but the actual automation of the marketing per person, right, like, on a on a one to one model. Mhmm. Like, an example of this is we have a system in a specific industry that customizes the products for them in advance and sends it to them, and all they have to do is say, yes, or I wanna edit this. Mhmm. Like, it already shows them what the product's gonna look like. They don't have to find us, find the product, upload their photo, put in their info, do all the customizing, get a proof. We just send them a proof before they even know that it exists.
[00:20:00] Corbett Drummy:
Right? Swag.
[00:20:01] Unknown:
So and we could do that on a one to one basis, right, for each person. And we can do it at scale. So
[00:20:08] Corbett Drummy:
What's what's interesting on this stuff too is the the early tests might work and then some might really miss. One that missed was Facebook and Instagram tested these AI influencers that were posting kind of bland and really weird inauthentic content. Like, it's something it's, like, really bland content for a human, but it's really inauthentic for an AI. But what's so funny about that is that these could work today, but what you'd wanna do is lean into the glitchiness and the weirdness of AI posting content. So probably the the most arcane and weird example that is, like, truth terminal on Twitter where some AI was hooked up to Twitter and just tweets really crazy stuff. And it's bizarre, but it's more viral and interesting than posting really bland stuff that is safe, you know. And so funny enough, like, Facebook and Instagram, they were testing AI content and it was received terribly. But I still think some AI influencers would work, but you'd actually want them to be kind of off the wall and weirder, and that's what would attract, like. Right. You're never gonna make a better TikTok than, like, the the current ones that exist today in the same style as them. It's just it's so dialed in with those how funny people are, how engaged they are. But you can make something different, and it can catch people's attentions through being just really out there. So, yeah, it's I think people actually need to take more risks and embrace more, like, the big companies, I mean. And for the the small and medium sized businesses, I think just adopting the technology is important, but, yes, it it's probably too hard. I wouldn't bother trying to carve new territory. I think just use these tools to optimize what you're doing today. You know?
[00:21:35] Unknown:
Yeah. I the meta just is one thing after another right now. They're just, like, flushing their platform down the toilet right now. I don't know what the fuck they're thinking. Like, honestly, like, I posted this this photo today, and it there's something to the effect of Facebook founder dies at age 36 of syphilis after saying misinformation will not be from our platform. You know? Like like, we're not gonna fat check it, but, by the way, he's dead. You know? Like, it and it's obviously untrue. Yeah. But, yeah, like, the whole thing is, like, okay. So they went, okay. We're not gonna allow virtual influencers, and they were like, okay. Yeah. We'll allow it. And then they're like, anybody who has more than one profile, we're gonna delete it. And if it's not a real person for you, we're gonna delete it. And then they're like, now we're gonna put our own ones of what we said where you're not allowed to do. We're gonna do it.
And then, of course, they were terrible because, like, no one wants to talk to AI Snoop Dogg. They actually just wanna talk to real Snoop Dogg. You know? Like and then they did the one where they were AI influencers that weren't real people, but they're pretending to be, like, average people. Yeah.
[00:22:45] Corbett Drummy:
That's the one I was referencing.
[00:22:46] Unknown:
Nobody wants to talk to an average
[00:22:49] Corbett Drummy:
fake average person. Like, most people don't wanna talk to an average person to begin with. How is there no one. Was about saying. I think that you're in the market for a start up to make a bunch of, like, kinda like, let let AI run loose, you know, not not like a Right. Not in that sense of I I mean, just you leveraging the the models that are out there today and, you know, already tested and saved. But letting them be more creative and do things differently than humans would. But using them to try to mimic the most average human is exactly what a large company like Facebook would do and is just, of course, gonna not work out.
[00:23:19] Unknown:
Yeah. It's just a complete misstep of, you know, who are our users? What do they want from our platform? I I don't know. I think it's obvious. You can take a small group of people pretty much anywhere, put them in a room with a couple of beers or a cup of coffee or whatever and say, what do you use Facebook for? And they're like, I wanna see events. I use Facebook Marketplace. I wanna know when Jim's having his baby. I wanna know when I gotta get to the birthday party. And, you know, like, they're not like, I really wish I could talk to a fake version of Bob. Like I know. I know.
[00:23:55] Corbett Drummy:
It's I think that, it shows why, like, acquiring startups when you're that big, like, a Facebook size is important because it brings, like, fresh blood and ideas into the company, but that's m and a has been a little bit cool off for various reasons, regulatory, economic, etcetera. But when you're just such a big company, it's really hard to take the risk that are required for, like, these new product things. However, you can take advantage of some things like the sheer scale of Facebook. Let it to be successful with things like, you know, they're the 1st people to really attack, like, the Craigslist with Facebook Marketplace, which is a great product. You know, people use it all the time. But, one thing They also they also said, you know, oh, there's been, like, 400,000,000
[00:24:33] Unknown:
AI uses on Facebook already, but they feed it to you without you asking. That's not that's not a use of AI. Right? That's that's the opposite of having people want AI. It's like it's I don't know. Like, it it's just the the opposite of getting people to use something is forcing them to have to use it. Yeah. Right? And that's what you totally do not wanna do. One thing that they are good at though and the these, like, mega companies, you know, TikTok, Facebook, you know, Meta as a whole, etcetera, Twitter and x.
[00:25:05] Corbett Drummy:
The algorithmic feeds will help with what you pointed out with the content singularity because we're already creating more content than we could consume, and that's only gonna be exacerbated. And that would be too impossible with following based feeds. But with algorithmic feeds, yeah, there'll be way more AI slop and bad content than ever before, but there'll be, you know, some content that's 10 x better just because the sheer content you'll have, the outliers that are real winners. And those algorithms are great at surfacing that. So I think everything shifted over. You know, TikTok really pushed it there. Others have tried it before to some success, but that really, everyone started shifting over after, you know, TikTok in the COVID era. Now it's it's complete. It's, like, all algorithmic everywhere. But that's really the only way you can handle the influx of content we'll see from AI. But, yeah, I think Facebook's, like, putting a thumb on the scale with, like, hey. They're force feeding these fake things to you, and that's they're kind of ruining their own algorithm there. I think things are gonna switch back.
[00:25:59] Unknown:
I think the pendulum is gonna swing back from heavy algorithmic content favored by the company Mhmm. To favored by the user. And, like, TikTok's algorithm originally was really good at saying, oh, okay. Well, you seem to like these types of things. You know? You seem to like gaming stuff, so we're gonna show you more gaming stuff. Mhmm. Right? But at this point, I mean, I go look at these apps, and there is very little that I am interested in that it shows me. Yeah. And what's happened is it's it's kind of swayed towards demographics, like really large generalized demographics.
And, you know, I get the the worst stuff being a 50 something year old white man. I get, like, do I wanna date girls from overseas, or do you want to, like here's here's 50 different Facebook groups that basically are just pictures of boobs. Like, this is just stuff that does not interest me. Not that I'm not interested in that, but it's not what I want on Facebook. Right? No. You know? So, like, it's just what I think is gonna happen is you're gonna have to have a way for people to start to make selections, like obvious selections, like do you want more of this? Yes or no? Yes. Right? And people will identify what they want, and then the algorithm can feed their preference that they have chosen, not a preference that we have given them a 100 different choices of things they don't want, and they have we pick the 10 that they watched the most,
[00:27:36] Corbett Drummy:
you know, which are still not things that they like. I I totally agree. The I think there there's some some that I've just stopped using recently because of the algorithms are so frustrating to me. Like, I've just stopped using Instagram, like, the Explorer and real reels are really frustrating to me. And sometimes it's funny how transparent it is where, like, if you, you know, like, one soccer video, it's like, man, you really love soccer. Here's a ton of soccer. And I've seen people gain their feeds. Like, the momentum is so sensitive. I I've seen people gain their algorithmic feeds where they just will, like, only, you know, images of people cooking with eggs and all of a sudden their entire feed is just eggs and stuff. And it lasts the short term, but the the frustrating thing is, like, when you might get, like, a funny meme like, I usually pop on. I just get, like, random general demographic memes and stuff. But then, it's not what I want long term, and I want to have that conscious choice of, like, steering it. Like, I wanna discover stuff in certain areas, and there's no way for me to push that. I don't get served much content in that area in general. And so I can't really lean into it. I don't get served much of it, and there's no, like, stated preferences. So I've And you can't search on Meta anymore because the AI system just tells you, like, a definition of what you're trying to search for instead of actually searching. And with your note about, like, Facebook itself, like, I'm sure that's just complete garbage over there. I haven't really logged it. Like, I probably pop it open once every 2 months and yeah. So, fortunately, I do usually see on my feed, like, oh, someone got engaged or had a had a baby, etcetera. And so a meaningful life update, but I I just have lost all utility from their products. So, like, I've I really have only used it once in the last few months for, like, maybe Google our Facebook Marketplace and once for, you know, just I don't know. I I just have kind of
[00:29:10] Unknown:
completely left. I would say for the first time in 17 years, I've bumped Facebook off the front page of my phone to a Yep. Secondary page because I don't use it anymore. I am liking Blue Sky a lot. Yeah. What's your experience been like on that? I've never I haven't did this ever since. Really, really good, actually. It has a good way for you to kinda start out figuring out who to follow. Yeah. And it is diametrically opposed to when you try to make a new account on x, which I just had to do for someone actually 2 days ago. And they're just like, do you wanna follow technology, sports, or Elon Musk? Like, these are, like, these choices at the beginning. Like, it's just it's completely made for they're like, what do you want us to show you that we're gonna decide on, not what do you want? Yeah. They so I I still probably use Twitter access my most
[00:30:01] Corbett Drummy:
used thing, but it's because, like, I've intentionally like, I get all my AI news from that, and I've intentionally followed people. Right. And every once in a while, I'll click over to the Explore feed, but the the the the amount they push Elon is so freaking frustrating. It's like I had enough of it so quickly and banned him, but it it still drives so much of the discourse. But I don't know. That that website is crazy, but I still need some outlet for finding some kind of new source, you know, like Right. I'm Well, Blue Sky Blue Sky allows links.
[00:30:31] Unknown:
Oh, that that's helpful. And they don't downgrade links. So Yeah. Somebody posts a link of, hey. Here's an you know, here's some deep journalistic
[00:30:41] Corbett Drummy:
story that we've done investigative journalism about this topic and there's a link to it, then people actually see it. Alright. Whereas on every other platform, they do not. You've gotten me down the the the marketing funnel there. I'm I'm now full like, I will trial it now. Like, I've heard it from enough people, you know. It's like influencer marketing. Like, you know, I've heard it from fam like, my uncle or friends, and now I will, cross the chasm and and try it out. Like, I tried threads, and it just didn't didn't really fit. So You guys can all follow me on there. I will. I will. I'll give it a I'll give it a try.
[00:31:13] Unknown:
My, my handle is author mattraus dothookdm.com. Cool. Alright. Well, you can actually use your own domain as well. So you could you could verify your account through your own domain. Yeah. It's just it's like starting up new networks is hard. It takes that activation effort.
[00:31:29] Corbett Drummy:
And but I do think also one I'd love to hear your predictions actually in your book, like, some that are coming up. But one is that usually with new ways of creating content, there's usually a new network. And so I I do fully expect that with the progress of AI, like, we'll see a new, like, family guy or Rick and Morty style thing that is really beloved and takes hold. And I don't know if that's gonna happen on on YouTube or, like, a totally new network. You know, it's I see lots of cool things created in our community. Yeah. And but, like, I I wonder what's gonna happen there, but there might it might be something in between a Netflix and a TikTok, like, similar way to stream like, get good content. But Hi. I don't know. But I I'm fully expecting with just, like, the revolution in in AI to breed a new way of consuming content. But I don't know if I think blue sky might be a slow burn that grows over time, but we'll probably see something rocket up from from 0 in the next, you know, year or 2 as it gets to be so easy to create and produce and and generate content that rivals, you know, kinda low budget, you know, animated series and stuff like that. Like, you can already create a family guy with with this now, but it's gonna be democratizing so easy that I think it'll we'll see new types of content being made soon.
[00:32:39] Unknown:
I think the first mainstream, like, single or small group project using AI is gonna be out this year. Probably soon. Like, within months, maybe weeks. Probably before my book comes out. It's just it's too it's too easy to do now. Right? Yeah. And with the editing tools and things that are out now, also with the ability to do stuff like like shoot live and convert to cartoon,
[00:33:07] Corbett Drummy:
you know, with, like, runway or something. Right? It's it's too easy to make it. So that it's a writing problem now. Yeah. We like, the thing that really can that came into me is when when I use style reference on our ltx studio product where you can make something and then with a single image, change the whole style of that project to be that image. You can, as a creator, get something on, like, I don't know, unsplash that or something that's, you know, you have the rights to or you can make something in your own style and then makes convert the whole project to that style and it pushes it out of that bland default AI, you know, kind of feel. But that way, you can make something distinctive and cool. And it doesn't need like, there's tons of hits that aren't fully perfect in their animation. Again, like, from Rick and Morty to anime or whatever, you can do that today. But it's it's Right. It's just hit like, I think the overhang of what tech can do versus what consumers have adopted is still pretty great and will only increase. But I think as that technology is be being adopted, probably once you can easily do it just from your phone, I really think we're gonna see an explosion of of content creation. It's gonna shift from, like, the, you know, the the Netflix is spending billions on content to these up and coming creators making full series. You know? Yeah. Yeah.
[00:34:18] Unknown:
And I don't think the the production cost is gonna be low enough to make something high enough quality that you could put it on Netflix or one of the other networks. And I think it's probably gonna come to the other networks first. It's gonna come to, like, Tubi or, you know, what was freebie kind of prime or whatever it is now, whatever they call the freebie version. There's so many names. I know. Roku and stuff like that. Right? It's where you can basically as long as you can pass the quality guideline, you can get on the network. Right? So as long as the people who watch it say, okay. This is good enough for our network that it's on. And then it's gonna be someone is gonna make something that is, you know, some one single creator or a couple a small team are gonna have some funny people mixed with some technically savvy people, and they're gonna put out something that's just gonna blow the doors off everything in Hollywood, and they're gonna shit their pants. I agree. Right? The great irony is that Creepy, which is one of the biggest flops of all time,
[00:35:16] Corbett Drummy:
is probably right, just wrong timing. Like, I honestly think there's a market for that stuff. Initially, TikTok ate it all up during COVID, and it's just wrong timing. You you didn't needed to go big blockbuster, like, HBO, Netflix style funded stuff, high quality, and, like, a little bit longer format or short form driven by creators from the ground up. But I really think, like you said, there's gonna be a single creator or small group make a real hit show or series or just, like, you know, comedy a couple short comedy sketches or whatever. And but I think that will end end up birthing a new platform followed by intense competition from the big ones.
[00:35:52] Unknown:
Yeah. I think one of the things that's probably gonna come out is some kind of streaming service similar to YouTube. Yeah. But it will probably be some sort of privately held group of creators who make a show or multiple shows, kind of like the Adult Swim idea Mhmm. Where it's it's all under the same umbrella. But the reason that I think it's gonna be done separately is because of the, like, the the supreme court turning over the FCC ruling in the United States about, you know, the Internet being I forget what the wording for it is. The net neutrality thing? Yeah. The net neutrality. So it will be easy for large cable networks to say, you can on your Internet package, these are the only streaming services you can use. Mhmm. Because they'll just block traffic to the other ones. Yeah.
[00:36:43] Corbett Drummy:
You can use. Mhmm. Because they'll just block traffic to the other ones. Yeah. I'm not really exposed much to the like, we're gonna have some downstream effects, like, you know, political and regulatory effects in the tech world. I'm not I'm not an expert in that area of, like, what we can expect. But I do think it's at least more open for innovation in the US market. I think Europe will lag behind a little bit with you know, they're already there's already a lot of AI products that are not rolling out in Europe because of certain regulations. And so Right. I think they're gonna lag behind a bit. It's a valiant effort to do certain things. And, like, because of things like g GDPR, it's changed how everyone approaches data. But it's it might be a little bit too early to have regulated yet.
You know, it's like regulating too early. You get stuck with the
[00:37:26] Unknown:
USB versus USB c or, like, inferior things for a long time. And but, yeah, I I I don't know exactly what's gonna happen in that neutrality and the downstream effects of all that. But Well, that's why I think things are gonna come off of kind of major networks and and be done outside of the current kind of large tech walled garden areas because they're not gonna be regulated. Right? It's the same idea as TikTok's new app, Lemon 8, I think it's called. Just Lemon with the number 8. Yeah. I haven't used it. It's basically just TikTok again. Like, it has some other features to it. Yeah. I had just downloaded it today. But so if they ban TikTok, they everybody from TikTok's already going over the other app anyway. Right? So they'll just, like, what are they gonna turn around and ban that app too? Like, I'm everyone that I was thinking of
[00:38:09] Corbett Drummy:
it. Like, they're gonna have that one day of ban, you know, the 9th January 19th coming up. I'm curious to see what'll happen there. But I would say in general, although TikTok's going through some of those, like, wavering a little bit, I think eventually they'll they'll be fine. But Yeah. I think they'll be fine too. But they I think TikTok and YouTube are best positioned to capture the influx of AI content. Right now old now. Right? Like, how long has TikTok been around? 8 years or so? It's like it'll be them versus a totally new network, but, they've got some advantages. But the the thing is But if you're 12,
[00:38:38] Unknown:
TikTok is what your parents use.
[00:38:40] Corbett Drummy:
Right? So you don't wanna be on TikTok. You're gonna be on whatever the next thing is. Yeah. Yeah. That'll be the next thing. It usually starts with people that have the tech in their hands and a lot of time, like high school and stuff like that. But I do think that the most interesting content right now is being shared actually on, like, Discord and private channels and stuff like that. Yeah. Like, the coolest AI content I've seen is is actually is kind of siloed. And so once that escapes and is being shared on feeds that could be, like, reshared, forked as projects, and contributed to, that that I think that's why I've been saying I think there's a market for that kind of new type of content, and I think it'll start to shift once we're already getting there. Like, I've seen some really cool stuff internally. But once you can just kind of be chat it's we talked about, like, GPT Canvas, and there's some stuff we're exploring. But once you can kind of just chat with these things as your creative assistant and it's like an agent, you don't really need the same interfaces as before, and they're they're going away they're going by the wayside. Like, I am not a coder, but but I can type in English and let gbt canvas code next to me and make a web app. And, like, you've mentioned you use it for writing when we were talking earlier.
But now I think it'll be the same thing about with video. Like, agents will do a lot of the creative production. Like, hey. I want this scene to look like this, this scene to look like that. And once that's kind of in our hands, like, mobile, I think that's when it'll explode, and that's when we'll I think we'll see this the surge of the new networks taking over.
[00:39:55] Unknown:
I think also, something that's definitely gonna be an up and comer is, like, an AI agent kind of, I guess, technically an agent that writes and creates and produces and exports its own show, right, where somebody builds the framework around it Yeah. And then it just makes shows. And once you once you have a system that can do it, right, then you make a better one and a better one and a better one. And soon somebody's watching this show and says, wow. This is actually a pretty good show. And then it takes off. Right? So I would say yes, but I think the intermediate step is gonna be just users prompting the agent to make edits. So I think that'll be like a as well. Yeah. That'll that'll last probably for a while. But we're gonna get there. Like, if anyone here has read, you know, Infinite Jest and there's, like, the the content that gets so addicting that it's hard to put down, and if you take away from someone, they get completely erased and just catatonic.
[00:40:48] Corbett Drummy:
I think there's that singularity of content where eventually, like you said, a AI agents will do do it without humans in the loop, and the content will just get recursively better and better and super addicting. And it might be that kind of a Wall E world where we're all in our VRs just in a kinda kinda bait, you know, on our floating hover beds. We like the glasses from the old RoboCop movie, you know. It's it's like a drug that is Who knows? Maybe we're secretly people are passing them around Yeah. Selling selling content. They'll be, like, in the back alley. They're like, you want some AI? Maybe that's our experience now is yeah, Matt. We're we're we're, we're actually in that right now. So who knows? It's right. We're on the way. It is the most interesting times to be both in I think the epicenter of all this progress is in the content world, so it's crazy to be in this space. But it's it's also while it's happening during our lifetimes, there's gonna be a lot of change and and problems, but there's just hopefully gonna be a lot of abundance and positive things as well.
[00:41:39] Unknown:
Yeah. Well, we gotta hope for the best because, I mean, it's coming whether we like it or not. So it's nice to be getting in there. Hope for the best, prepare for the worst. Right. And, honestly, I I have a pretty pretty cheery outlook on AI over the long term. Me too. So and, I mean, we got a lot of problems in the world. Right? We got a lot of stuff needs solving, and, you know, some of these problems are massively complicated, and and computers are good at that. So,
[00:42:06] Corbett Drummy:
you know, we'll see how everything goes. Corbit, we can talk about this all day, but we should probably wrap it up here. We've been already talking for, like, 40 minutes now. If people wanna reach out to you, what's the best place for them to find you? So I'm at corporate drummy on all socials. So you can just look at me on everything from, you know, Twitter x to LinkedIn. It's just at corporate drummy, Instagram, etcetera. And if you wanna check out the products we talked about, go to light tricks.com. You can see our suite of products there. If you're a creator, definitely try out LTX Studio for making these projects we mentioned. But, yeah, we'd love to chat with anyone if this stuff sparks interest. And, yeah, Matt, thanks for having me on it. Conversation's always fascinating, and we could keep going, but I think, yeah, no one wants to hear us continue to ramble. I think we packed it all in here. So thank you for having me. Let there be light. Yeah. I I like what you did there. You can read that Isaac Asimov story if you didn't catch the reference, the last question. It's a great one.
[00:42:58] Unknown:
Thank you for listening to the digital marketing masters podcast with your host, Matt Rausz. If you enjoyed the show, please leave us an honest review on your favorite podcast app. Be sure to check out Matt's book, will AI take my job, and have a wonderful day. Welcome to digital marketing masters.
[00:43:24] Unknown:
Today, we've got Corbett Drummy from Light Tricks with your host, Matt Rouse. With your host, with your host, Matt Rouse.
Introduction to Corbett Drummy and Lightricks
AI in Video Editing: LTX Studio
AI in Marketing and Content Creation
Building Products with AI: Challenges and Opportunities
Consistency in AI Video Models
The Content Singularity and Future of AI
Algorithmic Feeds and Content Discovery
The Future of AI in Content Creation
Regulatory Challenges and Innovation in AI
Conclusion and Contact Information
