48: Using Ai for Next-Gen Video Creation with Corbett Drummey

This podcast episode is sponsored by chiefaiofficer.com offering training and certification through the International association of Chief AI Officers. Interested in a new career or leveling up your value in the marketplace? Chiefaiofficer.com can help. The following is an interview with Corbett Drummey, Vice President of Brand Collaboration at Litrix. You may know Litrix for its popular Facetune app, where Corbitt is pioneering innovative tools like their LTX studio, which makes AI generated videos and storytelling accessible and efficient for everyone, regardless of technical or creative background. This episode's an interesting exploration into AI's ability to provide cutting edge creative skills to any of us, even if you don't consider yourself a creative. Welcome to Using AI at Work. I'm your host, Chris Daigle. Each week we'll be learning how today's business owners, entrepreneurs and ambitious professionals are getting more done with smart use of tomorrow's tech. Let's get started. Okay. I'd like to welcome Corbett Drummie to the podcast today. Corbett is with Lytrix and we've just had the opportunity to get to know each other a little bit better before we turn this on live. And I'm pretty sure pretty fascinated with the application of generative AI that he's primarily focused on, which is more of the creative side. That's not an area I'm just not wired to be the creative guy. So it's always interesting to sit down with individuals who are spending a lot more time on that use of generative AI because I just. It's something that all of us can benefit from and we don't typically tend to cover a lot on the podcast. So Corbett, welcome and let everybody know a little bit about first off, where you are, why you're there and what you're doing with Litrix.

Absolutely. Well, it's great to be on the show, Chris. Thanks for having me. And yeah, I'll start with background on me and the company where I'm at and we can dive into wherever you want to focus on. So I'm Corbett Drummi. I'm the VP of Brand Collaboration at lightrics. Lightrix is a technology company where I think most people know of us through our product Facetune. It's our most popular one, but we make a suite of content creation apps such as PhotoLeap and VideoLeap for photo and video editing. My background personally is I was the co founder and previous CEO of Popular Pays, the creator and influencer marketing platform. And two years ago we were acquired by Lightrix. So within the last like 612 months I've actually moved to focus on a brand new product which is one of the things we'll talk about today as well, called Litrix Studio. So Lytrix Studio is an AI video platform and it's our first product that's built from the ground up with AI and we. But we've integrated AI and pretty much all of our products underneath the light tricks umbrella. So today, you know, we can talk about our journey, like how we first started using AI in our products, how we use it day to day and then kind of the future of like these completely native AI products such as LTX Studio, what you can do with them, how they're kind of changing workflows, all the above. And finally I'm. I'm pretty passionate about it in my personal life, so I'm happy to talk about just tips and tricks from, you know, my own day to day work too.

Awesome. So I know that we had a different time originally for this podcast interview, but we had to change it because you are speaking right in London or where are you?

Yeah, I'm in London now. Kind of weird being in London on election day. But I voted before I left, so get out there and vote. I'm sure it's too late when people hear this. So yeah, I'm here because tomorrow we're doing two talks mostly about Lightric Studio. It's been a very exciting time because we released this product LTX Studio in March in beta. But then in August it's been open to the public and if you can think about it like a tool for rapidly prototyping video. But uniquely it's not just AI video as the outputs, it's a whole project. It's the video, multiple videos stitched together, narration, dialogue, sound effects and background sounds. So you really can go from 0 to 1 and a non technical user can produce a full movie like output in a few minutes.

What's the learning curve for somebody, somebody like me as I expressed, I'm not, I'm a non creative. What's the learning curve for me to be able to get quality output from LTX Studios?

What's cool is I think that first off I think Light Fix does a really great job in general with the UX and UI of that stuff of helping people go from concept to content quickly. But this one in particular, you can imagine it one of the ways you can initiate a project, you could put in a script, you could start from scratch, but you can also just kind of put in a prompt like you would chatgpt. And from that single prompt you can, you know, it'll suggest the characters and scenes and you can click, you know, confirm and it will create a whole project for you from scratch out of that single prompt. So really, in a minute you could go from just a blurb to a full movie project. And then the fun part is spending, you know, the next 30 to 60 minutes tweaking all the details. But yet even a non technical user with a few clicks can get an idea they've always had in their head out on paper and finally see it, you know, come to life.

What is the cost associated with being able to access that capability within the tools?

So some of the big news we released recently is that when it came out of beta in August, we made a version of it free for everyone. You get a limited amount of compute time, so about an hour. So imagine it goes a long way though. Like anyone can sign up today just at LTX Studio and there's that free tier. And imagine like if I'm creating a project, it might start with the keyframes and it will pull up a few keyframes, a few scenes, and I can generate the videos based on those keyframes and each video might be, you know, like 10, 6 seconds long or something like that. So if you think about it, that buys a lot of compute time. But if I export, you know, a five minute project, you know, I could do a couple iterations of that and still have plenty of time. So really a new user can jump in, create a first project. But if you're scaling multiple ones, like a, you know, marketer using this to prototype commercials or things, you'll want one of their business tiers. And there's tiers ranging from, you know, creator or business tiers that, you know, range from, you know, a few dollars a month to over $100 a month. And it's something we've been rolling out and iterating on based on the community feedback. But there's even like a custom tier for large enterprises with like a huge set of features that, you know, professionals might need.

Well, this is great. I'm actually, I hadn't checked the pricing before we got on. This is very accessible for anybody looking to do this. And the examples that are on the homepage, that is the type of output that I can expect to get from LTX Studio.

Yeah, yeah, yeah, it's really cool. So the, I think of all the things that I think are surprising about it is number one, most people are used to seeing AI video outputs as Just a single video. And the unique thing here is you're outputting a full project. So it's video, multiple videos stitched together. You can have, you know, 10 scenes if you want. And then there's also the sound design. The narration is really cool. We added dialogue recently so you can actually have characters talking to each other with lip syncing. Now you can upload a video of like a face in the emotion and even have that emotion drive the character's face motion on screen. So there's really these, like, powerful storytelling features that you can end up tweaking for, you know, a couple hours if you want to get the perfect shot. But I think the unique part is, is that most other video generation software is very slow. And here if you change something, like if I have a character who's, you know, a lawyer in a courtroom with a tan suit, I can type in red suit, or I can use a tool and delete something in the background, and it updates, you know, within five seconds, the keyframe and then a video. You can generate a Preview in another 15 seconds. And even when you're exporting the full project with all the videos together, it just takes a couple of minutes for the whole project, not just a single video. So I think what marketers have really seen, people in like the entertainment world, marketing world, they've seen that this is a really powerful tool for prototyping and getting your idea that your vision, you have in your head, getting it out on paper.

This is crazy, man.

Because.

Okay, let's talk about before this tool was available. If I wanted to do this, if I wanted to create the equivalent output of, let's say I spent an hour in ltx, I built something, I tweaked it. The output that I would get from that hour's effort, what would it have required? Pre generative AI, Pre AI tool.

The most interesting answer is probably that for most people, they couldn't have done it. And so, like, I'm a perfect example where I'm not technical. I'm both in terms of, like, engineering or design. And so I can talk about the teams I've been on where we have people like that with like, you know, skills around, you know, whether it's making static or emotion, et cetera. But someone like me, I wouldn't have the skill set to produce a video like this. Exactly. And, you know, I've had stories I've always thought of. Like, the first thing I did when I got this is I took a short story that was only three pages that I really Loved. And I had never seen anyone make it into a film, probably because it's too short. And I tried making it into one and it took me. I was going to play around with it for like 10 minutes and I ended up spending three hours playing around with it. And I would say it's hard to get. You know, one thing I have a lot of respect for is when you're trying to get everything to line up perfectly like dialogue with the most the transition into a scene, you can really go nuts like getting into details. And so I did spend, you know, several hours trying to perfect it. But I outputted something that it was alongside the vision I had of it. And it's something I could never have even gotten close to. I wouldn't even know where to start before I had ltech Studio.

So there was a technical restriction for sure. And let's say you had the budget and you approached a team to do this and they weren't enabled with these AI powered tools. What would you expect a budget to look like for me as an advertiser to create something equivalent?

Well, I'll give you an example. I'd say like the closest parallel I've seen is that, you know, a decade ago, before Popular Pays, before Light Tricks, I worked at an ad agency and we would create ideas for commercials for our clients. And we'd usually write a script. Like the copywriter would make a script. We'd have an art director make one or a couple photos and you know, they'd take like an hour or two comping up designs for just, you know, simple art direction of the of story. But it's the most like low fidelity way to communicate an idea. It's just a script and like a, A, A photo or two. Yeah, and maybe a photo per scene maybe. And so our storyboards would take the same amount of time, like hours to comp up because of the photoshopping of things, writing the script. And you know, now you can drop that same script into. First of all, you can use tools to help you write the script, but you can drop that script in and get actual video previews of your motion and scene like vision in like minutes. And so really it takes something. You would normally have to take a couple hours, like an hour or two just to get a static. And now you can get like a full rough cut and maybe the other like to punch it up a little bit.

Wow.

One time, you know, when we'd have to make like a more highly produced commercial, like maybe something we're doing for the super bowl or like an anthem spot. We would purchase stock footage and stitch it together. And so you're spending, you know, a couple like thousands of dollars on stock footage just for really basic stuff. And then there's no, there's rigidity there where you can't really tweak it. You're getting like a scene and a vibe. But here you can get the exact motion of like the, a spaceship coming down into the scene or someone walking along the street. You can add and remove objects. So imagine it's like this really interactive storyboarding. And so I've talked with people that have spent, you know, 10,000 plus dollars on a single storyboard that's rigid. And here, you know, you can get an unlimited amount of compute for less than that per year. So it's like a crazy, you know, a hundred x improvement over lots of people's normal workflows, whether you're in like the entertainment industry, the advertising, in this marketing industry. And so it's, it's one of the examples where it makes everyone good enough so they can do it. But people who are good, you can really get, you can almost realize your vision. I would say that it's not like it's not in a spot where you'd export it directly into a commercial. It's really a tool right now for prototyping. I've seen the curve of how much these video models are improving and we work on a lot of that technology in house ourselves. And I really think we should get used to the fact that in a year or so it will be there. We've already had some brands that have used this externally in ads, But I'd say 80% of people, 90% of people are using this for prototyping and, but more and more we'll use it for actual external content.

So I'm very interested in testing because I've got a couple of ideas like your short story that I've always wanted to see in video. And I'm not, I'm not trying to impress the world at large with perfection of our video. Would I be able to ship output if I wasn't? Okay, awesome.

Yeah.

Well then I can't wait to get started.

So a few cool things. I mean, I'll do like a little bit of a walkthrough of it because there's so much we packed into it. And personally I've never worked on a product that had more product velocity than this in the sense that the product and engineering teams working on LTX Studio are releasing things constantly every few Weeks we're adding, you know, major features or just like really like nice tweaks that make your workflows better. But as an example around after the launch of this in beta, we added some really cool sharing options. Some of them are, you know, you can share a project and collaborate it on it. So like you and I can collaborate on it as if it were a Google Doc so we can both be tweaking the scenes together. You can share it with permissions there. You can export it as a video file, like a complete video file with all the things packed together. One pro tip I would give people is do the preview first, which is like an animatic, so you can see static plus the sound, like narration, sound effects, et cetera. And then when the, when the timing of the scenes line up, then you export the final movie. But it's kind of like cheaper computationally, quote unquote to just create the animatic first so you can get the timing of your scenes right, then export that final project. And then underneath that you can even export like working files. So I can export something so you could start the product here and let's say you use some other editing software like Adobe or something, you can import the working files and continue editing it in that. So it's very like friendly with your workflows. And then finally one of the coolest things is you can export it as a pitch deck. So I can export like a. It builds a pitch deck for you in the sense of, you know, the title, the characters, the scenes, the color palette and design. It even has like, you know, the kind of synopsis underneath. And you can export this PDF to share if you're, you know, pitching an idea to someone.

So I know that your intended user is somebody in the creative field. Are. Have you had anecdotal experience with individuals who. You mentioned pitch deck and the first thing I think about was well, if I'm presenting a business concept and I was able to, because what I'm, what I'm always looking for is these tools allow all of us to stand out in some way by leveraging even whether it's the novelty of the fact that wow, I can make, you know, whatever the images move or the video avatars, whatever I'm doing. So I'm always thinking about how can I leverage these tools in a, in a non technical business use case, even if it wasn't intended for that. But in my non technical business use case, I'm going to stand out. I'm going to look like somebody who is capable with leveraging this new paradigm of technology. Are you seeing any users doing that stuff like. Like those uses yet?

Absolutely. There's a couple interesting examples there. One of them would be some people use this to as like a playground for development. So for imagine, instead of making a storyboard of a full project, maybe I'm just trying to get an image of a specific scene that I had in mind, maybe for a PowerPoint presentation.

Yeah.

And I'm. For example, one interesting workflow I saw is I had a visual I needed for a presentation. I use LTX Studio to start from scratch, and I just make a single video image, like video scene. And I don't need a full project. And I might be typing something like, you know, a floating orb in Times Square. Like, I don't know what the visual is, but you need a visual. And so I might just get an MP4 video and just convert that to an animated GIF. And so now you have a cool, like auto playing animated GIF for a presentation. So there's all these interesting ways where we've seen, like, I'd say a lot of people are making this, making full storyboards out of this, but many people are using it as a creative playground for trying many iterations of a single visual scene. And one thing that's wild, actually, we didn't even get to this. But I think my very favorite feature of it is what's called style transfer. And you can take a single image and change a whole project into a certain style. So let's say that you, your company has a certain style, whether it's a color palette or aesthetic, even something that's really hard to describe, like a pattern, you know, like Starry Night, you know, Van Gogh style. Like, you can take any, any type of visual design language, and with a single photo, it interprets it almost as you hired this amazingly professional artist to interpret all of your scenes. It is the most impressive use of AI to like, interpret the style to your existing projects. But you can get it kind of close and then kind of drag and drop your. Your stylistic reference there and get the exact thing you're looking for. So, yeah, I've seen people use it as a playground to test many different visuals and maybe they need it for, like I said, a presentation, something like that. It doesn't have to be a story you're pitching. It could be just a simple slide visual. That's probably one more example that's more like everyone makes presentations. Sure. Like, that's a really common example you might use it for.

So Most people on here have probably experimented with the image generation, generative AI supporting image generation. Video generation for me was something that I was like, it's not there yet, right? But it doesn't sound like. It sounds like I need to probably start to recalibrate my position on that because this sounds almost like ease of use of a, like an image generation tool. However, and all the functionality with being able to change the styles and things like that. Very simply, however, it's now resulting in motion video. So this is very cool.

I would say this like, I agree with your, I agree with your sentiment there. I think that I've seen some image generations that are photorealistic that can fool even an expert for video. I've used everything out there. I mean, I really have used every model possible and we've integrated the best technology we can as well as building our own for LTX Studio. I would say even some that aren't there yet, like, some aren't released, like Sora is not released yet from OpenAI. I think we can say safely that within a year and a half or so there will be videos that can kind of fool the a person who's paying attention. Right now it's best for prototyping, but you can still tell it's AI. But really the power has been okay, you're not going to make a full feature Hollywood film with AI today, but you can make a trailer in three minutes that can communicate your idea. And so I think it's like how people are using it today is really interesting, but we all should get used to the concept that this is coming. And I expect that, you know, there'll be a whole generation of creators now in the next few years using these tools who make the next, you know, like Rick and Morty or type of show where it doesn't need to be perfect and it can still be good. And then later you can make something that is, you know, it is quite compelling visually and really transports you into that, you know, cinematic world. But yeah, it's, it's really, it's really making. It's really democratizing who can be a creator. It's changing workflows. It's. It's hard for creators though, because they have to constantly learn these new tools and it can be really exhausting. As you know, we're constantly building new tools too. But it is exciting. And so personally, even though it's tough with the tides always changing in terms of what's possible, what you have to know better that than the static field that is never Changing. And so I think embracing that change is good. But yeah, for now I, if, if I were trying to get aware of what's possible, I would think if you've had a, an idea for a trailer in your head, you now who's non technical, can make that in, you know, five minutes and then maybe in a year you'll be able to make something that's pretty compelling that people would want, other people would want to watch in terms of like the artistic rendering of that video and how like high fidelity. High fidelity it is after that.

So just to get geeky and kind of switching subjects and I don't know if you can even share the information, but what is the engine behind the image generation? Is it flux? Is it something else?

Good point of flux. That's something we've, we've been playing with everything. I would say, yeah, the, I'll give you some background and yeah, it's, it's helpful to talk about like where you can use it today. For folks that are interested, we can talk a bit about what's under the hood, but we tend to use like we survey and try to use the best that's out there. And we also build some things ourselves. So for example, when we were first using it, we were using some parts of stable video for testing and learning, for just quickly generating things. We've also built some of our own technology under the hood. The first things that we released, for example, were places where we saw the market wasn't solving for. So even back in March, this is pretty unique. You could create a new character like at Ben. And in a scene you create, you can say at Ben is grabbing a cup of coffee. And it would keep the character consistency, which was previously something that was really hard to do between videos. And so we built things ourselves and I would say without getting into too much more just because hopefully there's some things we're announcing soon. We've, we've both been doing kind of pushing forward technology that's already there, like building on top of stable video as well as like really building things entirely new ourselves. And so if I'm being cagey there, it's just because we're hoping to, you know, talk more about that soon. I would say we're using the best that's out there. And our view is, you know, we can both be agnostic in the sense that even if we weren't building our own tech on it, we'd want to plug in the best of what's out there so that people can use us as like a platform, like a foundational platform to. To make what they want. And I would encourage other people, like other other folks that are thinking of using this in their own business. There's so much cool work being done out there with the foundational models, both on the LLM side and all the video side, that entrepreneurs and companies these days can kind of build on top of them in a way where it's like this nuclear power of free energy and you just have to build the piping and plumbing. And so even if you're just building interfaces, there's a lot of great work to be done and a lot of the challenges in the interface itself. But we think the fundamental problems and the foundational problems are also very interesting.

If you're enjoying this episode and want to learn more about how to start using AI at work, we've made it easy for you. For just $1, you can have full access to the Chief AI Officer community, which will give you additional training, custom software, daily Training, calls on AI tools using AI automations, getting more from your ChatGPT sessions, and the business of being an AI consultant. Simply go to chiefaiofficer.com insiders to accelerate your AI journey. Now back to the episode. How much has the technology, not just yours, but the technology of AI generated video output changed just since March when you launched this as beta?

It changes a lot. You mentioned Flux, which is an example where there are certain things where even over a course of a few months might be hard, and sometimes we build things for that, sometimes we leverage things off the shelf. But those two examples here are really pertinent. Like the one I mentioned, the character consistency, that's something that we spotted as a whole in the market, and so we built that ourselves. But another good example, and many people who've played around with Flux have seen this text is often very, very hard. And Flux can help with consistency and things like text. And so if you're a marketer right now, you know, you can make an ad of, you know, I want a scene of a car driving down Hollywood Boulevard. And you could do everything except you might struggle to put like a text on a billboard in that scene. Now you can, you know, so I think that it's. It's moving so fast that the one thing I kind of encourage people to do is if you had tested certain things, even like three, four months ago. Yeah, try them again. And so that goes with everything, like consumer tools such as ChatGPT, you know, I would encourage everyone to pay for the pro version. So you can try the newest models and same thing with this. I wouldn't turn a blind eye on it. Don't. In this field, don't write anything off because it kind of turns over completely every six months, like what's possible.

So I have a advertising background, primarily direct response digital was. I've done a lot of offline and direct mail and things like that as well. And one of the things that I learned early on that might seem counterintuitive to somebody who's not in marketing is that the pretty creative doesn't always win.

Yes, true.

And one of the things that we've been testing a lot with is so for us, our position on using synthetic output in our marketing is that it either needs to be indiscernible from human creation or we need to lean into the AI part, not try to fool somebody like the emperor wears no clothes kind of thing with. So I would be very interested in using this output in our creative. Not because it was perfect, because it was imperfect at some times. Because if it's perfect, to me it almost seems like noise. I'm scrolling, scrolling. Oh, another pretty person, another fast car, whatever. But if I see something that's like, wait a minute, that person has three legs, like I'm going to stop, right? And it's obvious that I'm not trying to fool anybody that this is human. But it sounds like the output that I'm able to get with LTX is, is it's not three legs, but it's not quite human yet. It's somewhere in the middle.

First off, you're so right about the like, I think it's the uncanny valley of. I think most marketers fall in the trap of being stuck there where you really do either want to push for like something that is like maybe used AI tools, but it's. But filmed for real. So you, you cross that uncanny valley. There's a. There's some models maybe in a year, you know, that will themselves create output that's. That passes the uncanny valley. If it involves humans, it's even harder, you know, because we're so good at perceiving that. But exactly like you said, the people that I've seen have the most success in exporting and using AI assets intentionally. Don't even try to cross that chasm. They stick with things that they, they embrace the glitches. One example would be like, I know the folks who do this, but we weren't responsible for this. But I'll point to it as a great example. There's like a semi Famous like Coca Cola ad for this reason where, you know, they're in a museum and it's a people shot for real. But then the painting comes to life as AI and it's, you know, reaching for a Coca Cola. But it's very glitchy. It's kind of like art deco or it goes to different art styles. And it's glitchy, but on purpose. And so it's kind of highlighting the real world versus this like artsy AI world. And I think that's been really successful. People leaning into the glitchiness. And even on a meme basis, like comedically, I've seen on TikTok and Reels, things blow up because they're leaning into that. Like there's a classic, you know, Will Smith eating spaghetti thing.

Yeah.

But even more than that, like there's these funny ones where people must be using other models like cling that are more prone to this. But they take like a natural looking thing, like someone walking and all of a sudden they're on a jet ski, you know, in the middle of a park.

Yeah.

And it's like that really like dreamlike super glitches that occur. They look kind of real from the, the video, but the scenery is so absurd. It's like surrealism that. That can be funny. And I think, I think either leaning into the glitchiness, like the Coca Cola example, or using AI tools to like add B roll but have the touch up of like, maybe it's you and I and there's something in between that's glitchy. Very cool to do either way. But I, I would, like you said, avoid the uncanny valley of trying to pass off like a. A weekly AI generated avatar as totally human. Like, that's not. It doesn't fool anyone and it doesn't really work that well. But if you lean into it, that's. That's great. It works great.

So kind of a curveball here. We are power users of, of like human video avatars. The preferred tool for us is hey Jen. We got a great relationship with them and we use them daily in creating content, creating advertising and that sort of thing. Is it how far are we away? And one of the things they do really well is if you are a good subject when you're recording your training data, it. It's hard to tell that it's.

Exactly. That's a great example of like, those are getting really good. I'd say like generating an AI character from scratch outside of like. Like, we have also added video that you can Have a character where you can upload a video to direct the character's movements that you like. Things like hey Gen, you can get pretty realistic when you have the video input or training, et cetera. Yeah. If you're generating a totally new character, it's, it's kind of hard. But yeah, they've, they've gotten really good. I've seen lots of avatars that get to be very, they're very realistic and that's another example of a field that's gotten better and better. Just like quarter over quarter.

Yeah. So where I was going to go with this is that if I've got those fine tuned avatars that I would like to use but maybe put me in situations and creative, is there the ability to merge that output from hey Gen as input in something like ltx?

Yeah. You know, you. First off, people have approached that problem in two ways. I've definitely heard that a bunch of times a lot of people create a project in LTX Studio and then export the working files to combine it with other assets. That's one.

Got it.

And you can also import like a lot of people have tools like midjourney as part of their creative process and they can import those images as starting keyframes and then animate them in Lightroom Studio. But specifically importing videos, that's definitely something that we've seen as a part of people's workflows that we're looking into. And in terms of a roadmap, we've been cranking out a lot of stuff to where whenever we see like something like that, which it's very interesting to us to tackle near term. But yeah, I would say you can imagine that that'll be a part of products and available soon. But for now most people are exporting the working files into some other editor or just starting with an AI generated image and using that to animate. But I think they'll all start, all will start to play well together because there's all these things you'd want to tap in the world, like for everything we're doing. There's really interesting things happening in, you know, video, sound, et cetera that you can imagine integrating as a stack in the future. And I think you mentioned that like you are probably one of the foremost experts in all the tools that are out there. You know, I kind of specialize in our world, but I'm sure you string a lot of these together naturally as well. And often early users are. There's kind of a clunky process of importing, exporting, but the more mature the tools get, the more it just gets easy to kind of string them together.

Yeah, yeah, we don't mind being early users of tools. We know that just over and over again we've seen that. Because I hear people, oh, I tried it. It doesn't work. When's the last time you tried it? Because I'm getting incredible output. Oh, I tried it a few months ago. It's a lifetime in the space. And speaking of that tempo, like, are you guys like, is there this, this ever present kind of anxiety about keeping up or staying ahead? Because things in this space move so fast and.

Yes.

Okay, like how do you guys address that in turn? First off, how do you calibrate that? How do you measure are we ahead, are we running with the pack or are we falling behind a quote unquote competitor?

Yeah, I mean, great question. It first off, I mean like I mentioned earlier, I'm sure creators and people in the space have this kind of low grade anxiety around the tools constantly changing and having to learn all these workflows. It's the same people, same way. For people making the tools, one thing I'd say is we have one of the most interesting, you know, internal slack channels sharing novel research papers and things. And so there's many cases where we go from, you know, I see the engineering team and research team go from seeing a research paper to testing and production or testing it quite quickly. And then sometimes, you know, we say, okay, this is better than we have in production. We want to, you know, be using this, this resource instead. But there's, since we have people working on research internally, they're really plugged into what's the latest in terms of what's possible with these tools. And so we're just constantly trying new things. There is a bit of this problem, like you said, where we might be close to releasing something and then we just judge that, hey, you know, it's, we want to push things further. And so we might adopt a brand new technology to like kind of leapfrog what's out there. So, so it's a constant kind of battle between what we can put out in production, what's around the corner, and if we think this new technology will give us a leg up or give our users better options in market. So we're just constantly looking around at that. But I'm not on the research team, but I'm definitely a beneficiary of just reading the interesting things that they're posting in that channel. So it's fun to see from the outside looking in.

So what are you paying attention to in the marketplace when it comes to the tools and technology?

I think from the product team. So I'll answer that in two ways. So we use or we use AI everywhere across Lightrics in all of our products from you know on Facetune you can do a one tap edit to you know it's one of our most popular parts of the app now where you know, it can simplify a workflow that you might have to like take a photo of yours and get it the way you want it to look when you want to express yourself online. And so there's these one tap edits on you know, Facetune there's video and photo effects in photo and videoleap and even in popular pays. Our creator influencer marketing platform, we released this amazing brand safety tool that can take a process that you know required hours of manual vetting for our brands and take make it just require minutes of the AI looking through a feed instead to analyze things. So we're using it all over. I think each product team is looking at what's possible in their field. So there's a heavy focus because it's used across our apps. There's a heavy focus on what's possible in video models like foundational video models. We don't do as much in sound. We still care about it for certain things but you know, like you said we're not like suno working on you know, music and stuff but we want to integrate that for our users where it makes sense. But we pay heavy attention to photo and video. On the popular pay side we've integrated it everywhere. From having marketers use AI to make briefs really quickly. Like on average the self service clients on Popular page writing a brief would take like the average time it'd be in draft was like days. And after giving them this kind of chatgpt like way they can put a blurb and it'll auto create a brief. It turned it into a process that was minutes and so we also had I think I it's been a while but we had like a huge increase in the amount of briefs being shipped as well just because it lowered the barrier to getting it done. But that was a project we did a while ago. But we're constantly look out of like what will be helpful for our marketers there. But instead of like video and editing tools we were focused on you know, how can we help brands and creators collaborate better. So it's really workflow improvements. But funny enough the kind of Like a great irony, so many people are worried about AI in terms of brand safety. Like, oh, will I be able. And will I be able to use this? Will my legal team give me the okay? And that's an interesting topic we can chat about. But the, it's funny that like the saving grace for brand safety might be AI itself because a lot of people have struggled with reviewing like staying on top of content at a mass scale in the age of social media. And now we've, we built tools that can, you know, look through just hours and hours of footage from now to years back and it can live monitor people in the future and, and it can look at not just the captions, but like audio in the video, the video content itself. And so from my, from my use of it, it's so much more effective than a human that I wouldn't even allow human vetting. I would only allow AI vetting for this kind of stuff because it can catch like a small, you know, text caption in a video from a year ago or this piece of an audio blurb. It's just superhuman. And so it's funny like we'll pay attention to different things based on what product it is, but in general, like super focused on video technology for our apps and photos, but also like workflow improvements such as the brand safety tool.

So right away, you know, like political season, you mentioned it was election day. One of the things that, that would be a downfall is if somebody, the picture from when they were in college and they were doing something stupid showed up. Right. Is, is this with. And it's the product you were just talking about. First off, would that be categorized as almost like social listening?

It is, in a way. It's even, it's even smarter than that though because like social listening historically has just been like text and most people don't really realize how powerful it's become where it's, it's not only that, but it's, it's cheap enough to where it's doable. Like previously crunching just endless hours of video would be too expensive.

Impossible. Yeah.

And now you can index it at a really reasonable cost and you can also live track it because it's so fast. But yeah, the like just like you said, now we're probably going to be past that era soon because any creator could use this tool and it's not only like, you know, of course you can find a picture of like a gun or something or like an alcohol. But it's so clever that an example is like it knows culture so for example, it, it knows that af, you know, means as, you know, F. It's, it's, it knows like abbreviations, it knows context. It knows things that might be indecent versus not. Like it knows the difference between a statue that, you know this, you know, a statue in the Louvre that might be, you know, a topless marble statue versus like a human by the beach. Like, it knows context in a way that's superhuman. And so it has found things that I find to be almost like eerily good at recognizing. First of all, it flags everything by default. And over time, like you could probably say, like, I only care about these things. Yeah, exactly. Like you could say that I don't care like I'm an alcohol brand, I don't care about alcohol flags or whatever. But it can capture nuance in language and audio, et cetera. So it's really impressive in that sense. But yeah, I think that, you know, we've had a lot of people have past posts come to haunt them and now with this tool, any creator would be able to look at their own account and, and, and curate it. And that's really good for them to have because we all post so much that, and not brands should have that ability, but we should be able to do that to ourselves too because it's our feed and it's great. It's gotten too, it's gotten too expansive for us to monitor our own content. Like who knows what, what we've been like. I've been in the social media field for 10 years and you know, we, a lot of people are growing up with, with this and so it's something that that creators should have as utility and, and brand should have as a utility both so they can like express themselves the way they want to express and, and partner with people that, that resemble their values.

This is interesting because being a business person, the first thing I think about is if I was going to make an executive hire for a high profile company, I would want to run this and see like, is there anything that's going to pop up before we make that hire?

It's so funny you say that I can't like say names and things, but like a sports team that was looking at using this for vetting people because you know, in the age of, you know, college athletes being able to use their, their social media feeds and just athletes in general making as much money from sponsorships as their salary, you can imagine heavy vetting of course on superstars just naturally because of public opinion, but there's a long tail of athletes in every industry that no one's really vetting until they become really famous. And this tool can, you know, make it happen so quickly. So it could be a utility for all types of things. You can also imagine one thing we're doing is talking with like really high powered, like global, you know, Fortune 100 brands who might want like kind of more custom things. Because you can imagine, for example, certain brands might be interested. Like if I make tools I might be interested in. Not just like typical things like swearing and alcohol. But is this work site safe? You know, maybe if I'm a food company, I don't want to show overeating. It knows context to the point where it's more than just we offer to everyone by default with these traditional flags, like anything that could be remotely brand safe or not, from politics to et cetera, it'll flag everything and then you can decide what you care about. But you can imagine really heavy power users over time kind of focusing this on things they care about. This might be unique to their industry.

So. And I know that you're preparing, you're preparing for a presentation. Is that what you're doing out there?

Yeah, yeah. So we're doing a few ones. There's been, I'd say, a huge growth in like the kind of AI summit landscape of people wanting to learn about this. So like your podcast as an example for podcasts, but lots of meetups which are really fun and we have been having a great time just highlighting LTX Studio to the world. So tomorrow we're doing both kind of a larger presentation as well as kind of a enthusiast meetup for demos in person. So that, that's would be really fun. But yeah, we, we also have a great series of events in the States. Like in a few weeks we'll have one in New York. So we're going to be doing regular community meetups and webinars, things like that. But if anyone's interested, follow our handle ltx Studio on Twitter. And we post about community things. For example, recently we ran a really cool event where we let people pitch us their idea and we gave them a huge budget for production if their idea won. So we gave them like a massive budget for actually bringing their idea to life if they made us the best pitch. So stuff like that, it's available on our Twitter handle ltx Studio and it's a really fun way just to track like the community comings goings. And there's also a great Discord channel where just enthusiasts are talking about, like talking about it Sharing results, things like that. But yeah, that's, that's for LTX Studio and for Popular Pays. Brands and creators can find us to, like, do sponsored work or to hire people for, for that work. But it's been a really exciting place to be in general because this has really transformed how we work personally and transformed our products. So it's been a really rewarding time to work on them here.

Anything planned for Austin?

You know, we just did a meetup for Popular Pays in Texas not too long ago. I think it was in Dallas, though. But we'll be at south by. So I guess if, if, if nothing else, we'll, we'll see you then.

Yeah, if you're in town, I'll get you some barbecue.

Right on. Yeah, I haven't been in a few years, but I had such a good time at south by last time I went and I, But I need to go in the off season, right? I mean, it's like coming to Chicago just for Lollapalooza. It's a little different than if you're there on the off week.

Now, we've got a great generative AI scene here, a lot of founders, a lot of builders, and I know that you, you've got engagements pending, but before we wrap up, I just wanted to get your take on who do you think is doing a good job in the space right now, regardless of what the modality is? Who's impressing you with, with how they're kind of leading in their category?

That's such a hard question. I mean, it'd be almost cliche to say OpenAI, but I have been such a fan. I've, I've tried everything and, but I just keep coming back to it. Like, you know, I've used Claude for things and liked it a lot, but I keep coming back to it for, for everything. Like, I use ChatGPT more often than Google Now. Like, 80% of my personal searches are through ChatGPT, both for, like, personal questions, workflow, things like, but also even random things. Like, I was traveling and I saw, you know, the airport book stand, took a picture of the books and said, can you rank these by, you know, Goodreads rating? And like, tell me which ones I like. I use it, like, so, so much that I, I have to say that one. But there's a handful of really fun startups that I see all the time and I think there's a. In each field, you know, like, we talked about Suno or Udio. Like, in each field, there's some really great standout companies in each like specialty. So hard hard to pick one outside of those, the gigantic ones. But we've, we've integrated and play with a lot of them.

Nice. Well, Corbett, I really enjoyed the conversation. I look forward to meeting you in person when you make it down to Austin.

Absolutely.

Enjoy your time over there and safe travels and anything that we can do to support what you guys are doing. I no kidding. As soon as I hit stop on this, I'm going to see if I can make that video I've always wanted to make right using ltx. So this has been been wonderful conversation, buddy.

Glad we got one more user. And yeah, if you, if you want you just, you can, you can message me if you want to help building something, but if anyone's listening, tag me on Twitter and I'm glad to help as well. Yeah, it's been really fun. Give it a shot. Let me know what you think. And just like stay tuned to it because we're releasing stuff all the time and we got some new things planned coming out in the next few weeks.

I'm excited to see that stuff happening. Okay, everybody, thank you so much. We'll see you on the next episode. Thanks for tuning in to Using AI at Work. Don't forget to subscribe for more conversations about how to use AI at work and a special thank you to our sponsor, Chief AI Officer for empowering businesses with AI education and training. Visit their website for a free AI Readiness Assessment and AI Strategy Guide to help you get started using AI at work. That's www.chiefaiofficer.com. so thanks to our producer, Evan Desaulnier for making this episode possible. Follow us on Twitter at the handle Using AI at Work and visit www.usingaiatwork.com for free resources to help you harness AI in your role.