Transcription
So the idea was really, really simple. You know, people were creating these AI images, but then they just don't know what to write. So we thought, okay, why don't we just create a community for AI artists where people can check out what other people are creating and then see their prompt? And then once we had this idea, we're like, okay, let's build a prototype.
If you think about food nowadays, you know, you need to go to Whole Foods to buy organic food. The analogy is that in the future, perhaps all the human-made videos will be like organic food, and then the majority of the videos you see will be AI-made.
Another strategy we started implementing later is programmatic SEO. That is actually how, for example, Pinterest and Canva really scaled up their SEO because you can create these templated pages. For example, we have a 3D art generator. So each of them may not give you a large volume, but then when you have like thousands or even tens of thousands of these pages, they really add up.
>> Welcome to Subversive, a podcast dedicated to sharing stories from the best consumer subscription [music] apps in the world. We'll bring you lessons for how to grow your consumer subscription business, including insights and inflection points that [music] led to exponential growth from leaders at category-defining companies and innovative startups. Let's get into the show.
My guest today is Koko Mauo, co-founder and CEO of Open Art, an AI-powered platform that enables anyone, including artists, creators, and brands, to turn ideas into images and short-form videos. In this episode, I talked to Koko about the company's early decision to pivot from image to video generation, about how Open Art has maintained its product differentiation in a crowded market, and about Koko's vision for the future of AI-native creation tools. [snorts]
All right. Hey, Koko. Happy to have you on the Subversive podcast today. Uh, we're going to be talking about your startup Open Art, which I've gotten to know over the last month, and I'm really excited about. So, to start off with, why don't you tell folks who don't know, what is Open Art and what was your path to starting this company over the last couple years?
>> Yeah, sure. Thanks, Phil, for having me. Um, yeah, so Open Art is a generative AI um platform for visual storytelling, and I'm happy to share more early stories because honestly, the narrative has changed over time, but then this is our most recent version. Um, and so people come to create their own stories, uh, mainly in the video format. Um, and um, so before Open Art, I was actually at Google for seven years. And this is actually another lesson I want to share with a lot of wannabe founders about like, you know, the the journey you might need to go through before you start your own startup. Um, but yeah, we started Open Art back in 2022, and um, so it has been, we have been around three years, and honestly, we jumped in at such a great time. Didn't know AI would be so big at that time. But yes, a lot of early stories I want to share with um founders or like people who want to become founders.
>> Yeah. And so we'll talk more about the origin story of Open Art in a moment and how you broke out in what is becoming a very crowded category, but talk me through those early days. So you're at Google, you've been there seven years. Obviously, Google's a very nice place to work and uh, clearly you had had some success there. What inspired you and your co-founder to to jump out and go all in on this startup?
>> So in my case, you know, even the day I entered Google, I wanted to do my own startup. However, [clears throat] you know, >> I just graduated from college at that time. One, I had no idea what I would start, even though I want to be, you know, an entrepreneur, a founder. And then second, honestly, like as a lot of immigrants may experience, like, you know, you just have a bunch of visa issues, and you know, once you got your green card, things become easier. Um, so that's actually these two reasons really made me feel, okay, let me grow more inside a company that had that can teach me a lot of things, grow my network, and, you know, one thing that was really, really fortunate that happened to me and my co-founder is that um, Google had an internal incubator, but unfortunately, it shut down in back in 2023, but uh, it started like 2016. So during the years um, you know, the incubator was there, we were able to like apply to the incubator. It's kind of YC inside Google. And we applied, and then our idea got, you know, accepted, and Google gave us funding and hack to pursue kind of like, you know, a semi-startup inside Google's incubator. And that was, you know, I was really, really grateful for the experience because it gave us the kind of the experience to experience what building a startup is like while having, you know, kind of like, you can say, like the security net of Google and having the support of the whole incubator um, and we were doing a short video platform at that time, which is actually quite relevant to what we are doing today, but it was um, nothing with AI, it was just like short videos because TikTok was still early at that time. So we learned so much from that experience. Oh, that's also how my current co-founder and I met with inside Google through the incubator.
>> Yeah. So, u, I think the goal was always clear, but it took me seven years to, you know, really learn, develop some ideas, and then figure out all of the logistics issues. So for any people who want who want to be founders or like, you know, encounter the same issue as I did, I just want to tell you that, you know, there is light at the end of the tunnel.
Yeah, I think there's there's this sort of curve where if you jump straight into starting a company out of college, it can work, and some of the most successful founders in history are college dropouts, but oftentimes you just don't have enough experience yet to have a high probability of success. But then if you wait too long, especially at a place like Google, you get comfortable and you don't want to leave. And so it sounds like you were in this nice middle ground where you you started your career at Google out of college. You learned a lot of good lessons and met some very smart people, including your co-founder, but you always knew from the moment you started there that you wanted to start your own company. And so when the opportunity came and it was the right window of time, you jumped out and started Open Art. And and so now let's let's move to Open Art and the origin story behind the company. So it's 2022. Uh, ChatGPT has sort of started to take the world by storm. Um, or maybe it hadn't even yet. Um, what inspired you to start Open Art specifically, and what was your initial vision for the product?
>> This is another story you probably don't expect because most people, you know, a lot of founders' stories you hear are very, very like visionary. Oh, when you start, you can see like in five years what you could become. And honestly, our story is a lot more humble than that. We actually quit Google January 2022, but we actually went through half a year of pivoting because our initial [clears throat] idea, which is another creator economy idea, just didn't work out well. It it worked out sort of, but then we just don't see it could become a unicorn. And then we started our half-year journey of pivoting. It was basically we would try one idea every week, and it's very different, you know, probably from a lot of stories you've heard, "Oh, I have this great idea when I was at Google and I quit and then immediately I I did this like successful startup that had immediately part of market fit." So our story is more like, "Oh, we two really want to work together and and then we think we had this decent idea and then we just left, you know, and it but it didn't work out." So it took us a year pivoting, and then one day, you know, we were at our normal weekly sync or weekly pivots and and then I was playing Dolly 2 on the side, and we were just thinking, okay, this thing is new. You know, it was August 2022, it was three months before ChatGPT's launch. So no one had the idea like generative AI would be a thing. The term generative AI wasn't like even a thing at that time. And so the idea was really, really simple. You know, people were creating these AI images, and but then they just don't know what to write. So we thought, okay, why don't we just create a community for AI artists where people can check out what other people are creating and then see their prompt? And then once we had this idea, we're like, okay, let's build a prototype. Uh, what's the most embarrassing MVP we could build? And turns out we just like crawled a lot of images online, like, you know, Reddit or Discord, and then showed their prompt, and then we had, and then we just posted on Hacker News. It's called like, um, the title was "Pinterest for AI Images."
>> Yeah, and then, you know, I think it was some luck, some good timing, and then it just blew up. It were like number one on Pro like on like uh Hacker News that day. I think a lot of comments were about how controversial it was, but again, like people had the pain point of like figuring out what to write. So from there, we got our initial traffic, and immediately we followed up with like, you know, good SEO and a good um creation tools. And luckily, Stable Diffusion also became open source at that time. So besides Dolly 2, we integrated with like, you know, other models, and then we started more like an aggregator of all the models. But yeah, a lot of things have changed since then, but that was the early story.
>> And so, if I'm understanding you correctly, it wasn't just that you were scraping all of these channels like Discord and getting the images people were generating, you were also including their prompts? And I can imagine how that was really powerful because in those early days, people didn't yet know.
>> In many cases, they still don't know what prompts they should be giving the LLM to get the image they want. And so, was that, it sounds like that was an important part of getting the early flywheel going?
>> Yeah, for sure. I think it is like, you know, even at a very early stage, it it was solving some sort of pain point, like people didn't know what to write for the prompt because ChatGPT didn't exist then, and then this whole thing is new, so people don't know what to type. So they need some sort of Pinterest-like inspiration for all the AI images. So that was our angle, and then, you know, MVP was easy easy to build. We built it under a week. So it, in our case, it's always about, okay, what's the most embarrassing MVP we can build to test out some theory?
>> And I'm always curious to ask, who were your early adopters? Were there a particular segment or set of user segments that gravitated to the product early on that you found were using it more than anyone else? You talked about AI artists, but even within AI artists, I imagine you have people creating all sorts of different imagery. And so, were there particular types of users that gravitated to Open Art in the early days?
>> That's a question that has bothered me since since day one. And the the true answer, which is probably not like very satisfactory to a lot of investors, is that the audience is so diverse. Like I've personally talked to hundreds of users on our platform since day one. And what I found is that it's just like very, very hard to like um summarize these users. Um, and and okay, if I really have to um, like say one type of users that I think I start, like early days, I saw some signals is um, for example, people who play D&D, like say, tabletop games, because those people, you know, traditionally they um, they need to find things online or have some certain artists to draw, but then a lot of them are um, it's it's just very hard to get these imageries when people play tabletop games. And then so you see a lot of those people coming. But again, like the group is so diverse, is that we couldn't say, "Oh, Open Art is is for like tabletop game players." It's not true. It's not representative. But there are a lot of them on the platform.
>> That makes sense. And I think with any creative platform, you end up with a more diverse user base earlier on. But there's still there's the stories of like, okay, when Instagram was born, before it was even Instagram, you had all of these hipsters in San Francisco who were using it because they loved the camera filters. Then eventually it became this network. And so for you, it sounds like you did have these pockets. And importantly, pockets of users who are part of a community because Dungeons and Dragons players or other tabletop game players, they're part of a community. And so if one of them discovers this product, now they're going to share it with their friends, and suddenly they're using it for their D&D games. And then they have other D&D groups that they're talking about it with. And so it can lead to this natural virality.
>> I want to share an anecdote. It was really funny because I constantly talked to users, and there was once I actually talked to two people who would um, who were in the same D&D group that both use Open Art, and then turns out one person shared with another person. And so, yeah, like this type of virality actually does exist.
>> Yeah. And it's so powerful in those early days because uh, even if the average user tells one or two other people, it gets the flywheel going. So, you start to get traction. You and your co-founder decide, "All right, we're going all in on this Open Art idea," and you start to get some early adopters, tabletop game players being one of them. Let's let's now talk about how did you get early acquisition flywheels going? You talked about SEO, so I'm curious to hear a little bit more around how SEO drove early growth, but then also word of mouth. Like, what were some of the early growth experiments you did?
>> And what worked? Like, what ultimately led to compounding growth?
>> Before I answer this question, one lesson I like to share with all the founders um or like people who want to do startup is that go-to-market is as important, or even more important, than than the product. That's why I feel like Phil, what you are doing, you know, helping companies grow, is so, so, so important. Because I think before you do a startup, you might hear you you might have heard wisdom, "quote unquote wisdom," that, "Oh, product is the only thing," or "As long as you're a good product, people will come." I think that era already passed.
>> I think, yeah, I think this is an era that a good product is the baseline, and then you need to have very, very strong go-to-market in order to um, you know, attract users, users, and and then you use a good product to retain them. But like, honestly, like the more the longer I do startups, the longer I feel, "Wow, like the go-to-market could be even more important than product." Because once you have users, you can iterate your product, and you can do a lot of things that when you have like, you know, 10 users, like you couldn't do much.
>> I think you're spot on. I mean, I think in the early days of the App Store, so circa 2009, 2010, you know, there were so few mobile apps in the App Store during the last major consumer tech boom that if you built a great product, people would just naturally talk about it and they would discover it through App Store search because there just weren't that many apps in the App Stores. But now, fast forward 15 years, I mean, everybody's building an app, everybody's low-coding at this point, which just means it's become way harder to break out from all the noise. And so I agree, I think go-to-market is very important. And so you entered, admittedly, you were one of the first movers, but still, you had Midjourney, you had Stable Diffusion, you had DALL-E, you had other tools to create uh, AI-powered art. And so how did you find a wedge that allowed you to break out and and what were some of those growth loops that helped you do that?
>> There are different stages of, you know, of Open Art. Let's say, you know, after the early story I told you, right? I think there was like, um, like 2023, honestly, we we didn't have much growth because um, we, I think one reason is that we're too lean, and then like we were not making enough progress, like and then we didn't focus as much on go-to-market um, but then during that time, the main go-to-market that really um, helped us was SEO. You know, SEO is something that takes time and also takes some luck, to be honest. For example, today, if you want to rank any of the top keywords like AI art, AI art generator, it is very hard. But then, you know, because we started early, and also because we have a very good domain name like Open Art, we easily ranked like, you know, the top keyword for um, top result for AI art. So that's probably one suggestion I would give to founders is that think about like think of a good domain name that could really help boost SEO for you. Yeah. Um, so so then like, you know, the SEO also consists of normal SEO, which means you need to make your landing page or homepage, each feature page optimized so that they could get indexed. Another strategy we started implementing later is programmatic SEO. That is actually how, for example, Pinterest, Canva really um, scale up their SEO because you can create these templated pages so that like you can target like niche. For example, we have, I don't know, like there are too many pages we've created, like um, 3D art generator or AI like D&D generator image generator. You know, all of these things. So each of them may not give you a large volume, but then when you have like thousands or even tens of thousands of these pages, they really add up.
>> I'm smiling because this was exactly Quizlet's playbook. So for those who don't know Quizlet, Quizlet's one of the largest edtech companies in the world. And the way that it broke out was through long-tail SEO search queries targeting digital flashcards, which is the atomic unit of content on Quizlet. And we called it programmatic SEO. It was literally the same game. So I'm curious. I mean, partly because I know this world well, like what were some of the very specific tactical things you guys did to make programmatic SEO work? For the other growth marketing product leaders out there listening to this podcast who want to build an a programmatic SEO strategy, what are a few of the most important things to get right?
>> So even though you have a template, like each, I think content on each template should still be, you know, differentiated enough and then high quality enough so that Google actually doesn't penalize you because if you have too many duplicate pages, Google will think um, they are duplicate and then it actually hurt your domain authority. So it is very important to make sure like each page is differentiated enough, especially the content. And the second thing I would say is that find really like um, like how to find these keywords. One of the um, well techniques is to look at um, whether you have competitors that already, you know, did the programmatic SEO to look at what keywords are they ranking, and then you could try to rank, you know, similar keywords. So that's kind [clears throat] of um, another shortcut you can do. And and three is just to make sure these pages are optimized, you know, because everything, performance, like uh, like, you know, like image size, all of things, all of these small things count, so make sure your page is very optimized.
>> Yeah, this sounds so familiar. I mean, at Quizlet, it was a lot of the same stuff. It was making sure that we didn't have too much duplicate content, or if we did, that we were de-indexing it so that we didn't get penalized by Google. It was making sure that the creation flows for creating content on Quizlet were as quick and easy and seamless as possible because that just led to more content creation, which led to more content being crawled and indexed by Google. Um, and then it was especially in international markets, which was one of the big things I focused on there. It was figuring out where were the pockets of specific subjects of study times, particular countries where we didn't have content coverage, and then finding ways to incentivize students and teachers in those markets to create the content to fill in those gaps. And I'm guessing that you probably adopted a lot of those same techniques to get your early flywheel going.
>> Yeah. And then we also did um internalization of all the pages after, you know, where um, I think today we probably have thousands of pages, and then we actually um, find the top um, like countries of our top subscribers and then translate those pages. Um, I think that is also Yeah. Helping.
>> Yeah. Yeah. So there's translating the content, and then I don't know, you have all done this, but having country-specific subfolders, getting other websites with high domain scores to link to you in those international markets, all of these are tricks for like jumpstarting international growth through SEO.
>> Yeah, exactly. So I I do think SEO is a continuous effort, and then it takes time, but once you know, once you have a solid foundation, it continuously drives traffic. Though even nowadays, to be fair, a lot of people are talking about G what is it like GSO, like now?
>> Generative, generative engine optimization?
>> G, yes, yes, like, you know, because people are starting to use chatbots um, so that's one thing I think everyone should you know, look out for, that could be a new good go-to-market. But to be fair, if you're good with SEO, that's definitely helps with your GEO.
>> Definitely, yeah. I get that question all the time, which is, "What do we need to do to rank for GEO?" And the answer is, in many cases, a lot of the same things you would need to rank for SEO, but obviously, there are some differences. So, okay, so you've launched the company, you come together January 2022, you spent six months pivoting, you find Open Art. Now you've got early traction, and initially it was images, but then eventually you pivoted [clears throat] to video, and there was this idea of the one-click story. So tell me a little bit more about Open Art's evolution from image platform to video platform, and then consistent sort of storytelling platform.
>> That is such a great question because again, I think this is another very um humbling story. Like last year, you know, like 2024 um, we actually grew from 1 million to 10 million. So from outside, it looks so impressive, but then from inside, I was personally very, very anxious because u, all we were building was like, you know, >> it's a it's a workflow for AI images. There's like, okay, image generation by aggregating all the models, there's image editing, which consists more of like um, proprietary tech. However, you know, it's just not as differentiated. Whenever I talk to investors, the top question they ask is that, "How do you differentiate?" You know, and you could argue, but then from, you know, deep down, I also think, "Hmm, I feel it's hard to differentiate." And >> yeah, and then so I, at one point, I thought, well, you know, I I didn't want to, you know, end up being another doing another startup that got acquired because That's what happened to the previous startup. We did an incubator. It got acquired by Google Search, you know, which which was good results for a startup at Google's incubator. But again, like I thought, this is such a gold rush time that like if we don't do something big, the opportunity just maybe forever missed. Um, so [clears throat] I thought really, really hard about, okay, what will our long-term strategy be? Like, think deep also about what I'm personally interested in. And luckily, you know, Open Art has like three, four million monthly active users. So, it's really easy for us to access our users to like verify our thesis and so on. Um, so the the main thesis I came up with was around visual storytelling, which is around, you know, at that time, a lot of um, the AI platforms, including Open Art, was so focused on individual assets generation, generate this image, that image. But what we've heard from users is that consistent character is something a lot of people want. And when you dig into why people want consistent characters, is that they want to tell stories. Um, and by story, I mean, it's a very broad concept, you know, uh, like a film is a story, like an like, you know, a short TikTok video is a story, even an ad is essentially a story with a call to action to buy something. And then you started looking at their workflow, and then and you'll see like, a lot of websites are just not built for that because for storytelling, you want the characters, you want the narrative, and you want audio, and all of that. And and so That's when we start thinking, okay, if we focus on storytelling, can we be more specific, build more features around storytelling, be more differentiated? And I also believe that if we focus on storytelling, there's a path for us to toward transition from a only creative tool to a more platform level um, so that's when we started doing the transition. And once we realized, okay, video is actually the best format for storytelling instead of, you know, a bunch of images. And luckily, video models are also getting huge, huge like um improvement early this year. Um, so with all the things combined, we started pivoting to videos and started pivoting to like this stories.
>> So as you pivoted from images to video, that makes a lot of sense. Like it's a harder problem to solve, which means it's easier to differentiate your product solution. What did that look like? What, what were some of the key changes that you made as part of this transition to video and storytelling?
>> On the one hand, um, the transition is actually may not as drastic as it sounds. The the really interesting thing is that when people make videos, the most common way for people to make videos is to start with an image. Like image to video is actually way more commonly used than text to video because it gives people more control, give like it's actually economically make makes more sense. Um, so so the image part of Open Art actually serves as a really good foundation for the video part. But in terms of changes we had to make, there are like actually a number of changes. One is like, you know, experience-wise, like um, how do we um, make people um, you know, how how can we assist people transition from image to videos, especially like um, if and honestly, some people even didn't realize they need videos or realize they have the need of stories. So how do you upsell to existing users to this new offering is one thing. And another thing is about new go-to-market. We started exploring other go-to-market that could work for video users. And third thing, which is very relevant to what you are doing, Phil, is about subscription. At one point, we realized that, "Wow, with our current subscription," because one long, long time ago, we had infinite subscription, which is like, you can have infinite credits. Once we introduced videos, we realized that we would be bankrupt if we don't change our subscription. And we made the hard call that, okay, we're removing this change, and then we're going to put a hard limit, but we're going to allow add-ons, so people can, you know, purchase additional credits. That was a very hard change because you can imagine how unhappy people were, you know, like complain on Discord, and then even like send us some like, you know, hate speech, all of that. But one, I think that's the right thing to do for the business, and second, in spite of all of this hatred, you realize that people just purchase despite of their dissatisfaction.
>> This is one of the things that I say to reassure my clients from time to time, which is if people care enough about your product to tell you how upset they are about a change you made, that's actually a good sign because it means that they're emotionally invested. Now, obviously, you want to do everything you can to make your customers happy, but every once in a while, as a startup, you need to pivot and you need to make changes. And if people care enough to complain, then that means they're probably going to stick around as long as you treat them fairly. So, let's talk more in detail around what was your subscription model before, and what is the subscription model you offer now? And as much detail as you're willing to share. It sounds like before you had this infinite plan, which made sense in the context of image generation because the underlying compute cost was so much less significant, but with video, >> I can only imagine how much cost you could rack up in terms of underlying LLM costs very quickly. So, what does your subscription plan look like now? What different tiers do you offer? And then how do AI credits factor in above and beyond those foundational subscription tiers?
>> Yeah, so we we had three tiers, right? Well, it excludes the free tier. Well, our free tier is only some trial credits, so it doesn't really justify a free tier, but we basically have three um tiers, uh, three subscription tiers. Um, there's Essential, which is like, we recommend to anyone who wants to just start, and then there's like um Advanced, which if you need more credits, and then Infinite, sorry, it's Infinite, but which which you actually have like the most credits. And honestly, the main differentiation between tiers currently is just credits. We tried to feature guard things, and then we didn't find it work very well for us. Instead, like people run out of credits, they will just upgrade. And the the most important change we did last year is that we have this additional add-ons because, you know, when you go to Infinite, what if you want more credits? And and then so we can have these add-ons where we can purchase more, and it actually worked out pretty well for us initially. We can say or we said, "You can buy 20 credit packs or even 10," I don't remember. And then some users complained, "No, I want to buy more, can you do more?" And then we just like expand the limit. So yeah, I I definitely felt this like add-on like incentives works pretty well.
>> And I assume that you are offering a subscription because of the recurring revenue effect of that, right? You could just offer AI credits, you you don't necessarily have to offer subscription at all, but the benefit to a subscription is it ensures you're getting a recurring revenue stream and it builds more of a habit with your users. Are you discounting the per-credit cost of the subscription relative to just buying AI credits all cart?
>> You actually cannot just buy the AI credits. Like we require a [clears throat] subscription for you to buy AI credits. So >> yeah. Yeah. The thinking here is that, you know, like we want users to keep coming back, you know, and then if they just buy the credits, sometimes they just like, you know, leave it there and then they forget about it, and then they never come back. So I do think um, we wanted people who are more engaged.
>> Yeah, that makes a lot of sense and it's obviously better for the business, giving recurring revenue, but it's also arguably better for the user because they're building more of a habit and getting more value out of the product. Uh, and it makes sense that you have to have a subscription to get long-term access to the platform at all, but then if you want additional usage, you just pay for credits on top of that. How are you balancing organic versus paid acquisition at this point? Have you started to spin up a significant amount of paid acquisition, or is most of your subscriber acquisition still coming through word of mouth and SEO? [clears throat]
>> I would say we have a mix. So um, I know a lot of investors have bias toward paid marketing because, you know, it's not organic. Um, but in our experience, you know, like and and I talked to a lot of like later-stage founders um, I think one um, one consensus that paid marketing is something that can be scalable, like, you know, if you can really scale paid marketing, like it's actually not not like it's it's good to have because you have a way to scale it up. Um, but on the other hand, I also do think um, like every startup should control the percentage of their paid marketing because, you know, you really don't want to be too reliant on paid marketing so that if somehow, you know, things happen to like to ads, like there's policy issues or something, then you get screwed. So our top priority right now is to diversify our go-to-market. We have paid marketing, but we also have like um, you know, organic growth. We have social media. In fact, social media is one channel that we've been um, really putting a lot of efforts. Like our YouTube channel, for example, grew from like earlier this year was like 30k, and then now it's almost 200k. Um, and and so we found a lot of benefits with like social media because it's one, it's free [laughter], unless you're like paying some labor, but then it's essentially free, and then second, it really builds the relationship between you and and the viewer because they see you as [clears throat] a person, and they get a lot of like educational value from from this. And if you do it right, like YouTube, it's essentially SEO.
>> Yeah. No, that's a great point. There's definitely some overlap across some of these different channels. And speaking of which, one strategy that I've seen more and more consumer apps using recently on TikTok and Instagram, particularly, is finding influencers or micro-influencers to create UGC content, but then doubling down on the best performing content with paid ads just to give it more reach. Given that you are a creator platform and you have all of this incredible video and storytelling content that's being created, is that something that you've found to be successful?
>> We actually did it several times with um, like on YouTube. Um, in our case, I don't think we've fully cracked the code of that that loop yet. But yes, I I think you're totally right that like combining this organic with paid could result in something good. I think we haven't cracked the code yet, but this is something we want to try.
Yeah, I mean, this is consistently what I hear with TikTok in general is it's a tough nut to crack and it's much less predictable and reliable than Meta or Instagram. But when you really figure it out, or if you can find a a content type or or a creative concept that works, then the upside is much less capped. Uh, and so you have to take this sort of portfolio approach where you you create a lot of content, and a lot of it doesn't work, but some of it does, and and then that drives a lot more upside. So let's talk a little bit more about growth metrics. So obviously ARR is a very important one, and you mentioned that you grew from 1 million to 10 million in ARR last year in 2024. So congratulations on all the growth. Obviously, things are going well. But underneath ARR, what are some of the most important growth metrics you look at and why do you focus on those?
>> You know, obviously, I think revenue is very, very important. And I think besides revenue um, other metrics, I don't know if they count as growth metrics, we really care about. One is retention. Um, because, you know, I I I regard retention as a growth metric because, you know, the more people you can retain, it's essentially growing your user base too because they don't leave.
>> Absolutely.
>> Yeah. Otherwise, they just leave. So um, retention is something we really pay close attention to. Um, so and retention, and then second one is um, I would say again, I don't know if it counts as gross uh growth metrics, is um, gross margin. I think it's particularly important for sustainability of a generative AI startup because if you are at negative growth margin, you know, then either you have a lot of funding, or like you don't care if you survive in a few years and just waiting to be acquired. But if you really want to build a long-sustaining business, I do think you need to care about your gross margin.
>> Yeah, I love both of those answers, and I can tell that you're a product-driven founder just based on how you answer that question because if you asked a lot of CEOs, "What growth metrics do you care about?" They would say things like ROAS, return on ad spend, or, you know, LTV over CAC ratio. And those are important metrics, but at the end of the day, I love that you started with retention because the reality is, if you have subscriber retention rates that are high enough, >> that can make up for almost everything else. So, if you have a strong foundation of people who can't live without your product and are continuing to pay for it, then over time, that is the foundation for every really successful consumer subscription business. I just had Demetri Gerski, who's the co-founder and CEO of Flow, on the podcast recently, and he talked about this, like subscriber retention has been the absolute keystone to Flow's long-term success, in part because as a women's health and period tracking app, like their retention is extremely high. It's it's a use case that every woman is going to have really from the moment they hit puberty through most of the rest of their life. And so they've built the whole growth strategy and the economics of the business around really strong subscriber retention, as well as having healthy gross margins. And that goes back to your point around how you're monetizing with AI credits and making sure that power users aren't running up costs that are so high that they're actually unprofitable. And so I love that you you said retention and gross margin. I think those are two of the most important growth metrics. Even though oftentimes they're not the first ones people think of because they're so focused on acquisition and trial start rate and subscriber conversion rate. At the end of the day, it's actually those deeper funnel metrics like retention that matter most.
>> Yeah. And, you know, one thing I realized about retention is that it's compound. You know, if you have retention right, that each month your cohort retention will like, it's a compounding effect.
>> Absolutely. So, uh, it sounds like Open Art is off to a great start. You are now a few years into starting the business, really just a couple years into the start of Open Art itself, and you're on this incredibly fast ARR growth trajectory. Um, let's talk about the future of Open Art. So, to start with, I guess I'm curious about what's your view on open versus closed models in the AI content creation space, and how do you think this market is going to evolve over time?
>> Well, we benefit so much from open-source community and open-source models that we, I'm definitely very, very like advocating for open source. And honestly, I think I've never seen open source like um, growing as fast as now. Like it's almost like if you have a closed closed-source model or some tech, and then in a few weeks or in a few months, like open source will just catch up. And and, you know, for a startup like us who don't train like models, you know, open-source models give us the ability to really customize things to our own needs. For example, like we've built for like, in our case, right, we of course integrate APIs with all the closed-source models, but then if we want to do something more creative or like tailor to our own needs, then open source gives us way more flexibility and freedom to to like to do that. Um, so I like, you know, for our own benefits, I really hope like open source will keep growing. And um, and I also think that like open source is definitely the right thing for the, you know, the progress of all the technical breakthrough.
>> Yeah, I mean, that makes a lot of sense, and I think we're hearing more and more cases of open-source models pushing the frontiers of the market, both in terms of the capabilities of AI, but also in terms of the costs, the underlying costs of the LLM.
>> Um, so that's really exciting to hear. How do you think the market for AI content creation is going to evolve over the next couple of years? We've already made this pivot from images to videos and storytelling. Like, what do you think are the most important >> characteristics of the companies that are ultimately going to succeed in this market, and how can startups like Open Art continue to compete in the long run against the big guys, the Googles, the Open AIs, and the Anthropic's of the world?
>> I've been thinking a lot about it since Sora 2 came out. You know, Sora 2 is really the next level compared to a lot of previous models. It really synthesizes audio and, you know, the whole video thing really, really well. And now you can create this like almost um, you know, like there's an app called V uh, no, no, sorry, Vine. Vine, there's an app called Vine in the early days, and it just really reminds me of like Vine, like all of these like 10-second videos, and then you scroll, and then you wouldn't realize you are scrolling a generative AI feed because they look so real. And and for us, you know, our vision is to do visual storytelling. And I definitely see like as the models getting more and more um competitive and then more advanced, perhaps they can just create like the full story directly from the model. Then what what do we need to do, you know, as an application company? Um, so I think there are a few things. One, um, like think about like, for example, like the um, the needs that for example, these foundational models couldn't serve. As example, you know, if you want to create longer stories, you probably need more controls, and you probably need all of these fine-tuning that like just a model cannot directly output. So giving people more control in like in um, in logo format is one thing. And then second is that besides product and tech, what are other other values you can provide to your users? As example, one thing we're pursuing is around like, let's say um, if we help people create their characters, right? Then by itself, it's just a character. However, you know, like one of the most valuable IPs in the past decade is Mickey Mouse, right? But then in order to build this successful IP, you need a lot, like you need legal protection, >> you need the whole world around Mickey Mouse. He has friend Donald Duck, you know, like I think there are just a lot more from the tech level we can support to the users. So as example, like today, if you use AI to create a character, can you legally copyright it? Like I think it's a it's an interesting question that's still in the gray zone, right? Like some company has to figure it out, like, you know, if you really create the next Mickey Mouse with AI, like you got to protect it, for example. So like I think there are a lot of values we can provide to these creators besides like product level, like again, help them monetize, help protect their like, you know, copyright and so on.
>> Yeah, that makes a lot of sense. And it reminds me a little bit of I worked at Faire for a period, and Faire is a marketplace that connects brands and retailers selling physical products. And so the tip of the iceberg is product discovery, right? If you're a retailer, you want to discover interesting products from other brands and fill your physical brick-and-mortar um space or your online website with those with those products. But then behind behind that, you have all of these additional services that these creative people, these store owners need. And that includes CRM, it includes financing, it includes operations and logistics, shipping, fulfillment.
And so I love that you're already thinking about all of the additional SAS tools you could build for a video or story creator because if they are really able to achieve some level of success, then they're going to need help protecting their IP and selling their product. Um, so that makes a lot of sense to me.
Another thing, you know, fair did well with the network effects. You know, you have the because it's a marketplace, you have the seller, you have the buyer. So I think that's one thing, you know, we also want to correct somehow, create a platform with network effects. You know, I think that's what Star is trying to do, like creators and consumers. Um, but then, you know, it's how do we differentiate if we build a platform that's, you know, another question for us to think about. But in general, if you can build network effects, that's the gold.
Yeah, well, it's great that you started as a network product from the beginning and you really built it around these communities of creators. And so it's in your DNA and it sounds like there are plenty of opportunities to lean into that going forward. So you've already started to talk about this, but imagine we fast forward five years. What is your vision for Open Art in 2030? And to the extent you can share, are there any particular new features that you're really excited to launch over the next three to five years to support that vision?
You know, Phil, I I like this question, but I would say, you know, five years is a long, long, long time. space like if you like, you know, I like to predict, but then a lot of a lot of times it's like highly impacted by, you know, like, um, all of these model improvements and then even like the market and so on. So I think five years maybe too long to predict, but I could definitely tell you like in the next one, two years.
Sounds great.
Yeah. Um, I I think, um, one, we, um, like I think we really want to focus on the visual storytelling component because we think [clears throat] it's just so inherent in human nature that like people want to storytell and Jai is making it like basically possible the first time that like average people could direct a Pixar-level of film. Um, and so we're very, very excited about the potential and want to democratize visual storytelling. Um, and right now, you know, Open Art is a creative tool. Um, but both me and my co-founder, we are very like, um, like at our heart, very community-driven people. Even with the last, we did, you know, it's a short video platform and we serve a lot of lifestyle short-form, um, content creators. And even though, you know, we didn't get to the scale of, let's say, TikTok, but even with our scale, we we really felt the power of community. Like even when you had a small community, and if you can get the flywheel running, it is very powerful. So like our vision with Open Art is that can we be the platform for visual storytelling, which means on the one end, people can like, you know, create these stories, but on the other end, people can really experience like visual stories in a new way, like perhaps like be more interactive, more playable. Um, so that's where we want to see where Open Art is going.
Understood. Yeah. So, it really is centered around this idea of storytelling, which has been such an important part of human history from the very beginning. But as technology has evolved, there have been more and more advanced ways to tell stories. And so, you're putting the power of storytelling in the hands of everyday people, uh, who, you know, previously had to go watch a Pixar movie to watch stories of that quality. Now, they can do it themselves. And then just making the stories richer and more immersive. It sounds like that's that's the primary thing you want to accomplish over the next couple years. And then we'll see where we are in five years because that's just too hard to predict.
Yes, [laughter] but that's a great summary, Phil.
Perfect. All right. Well, let's get to our lightning round. I'm just going to ask you a series of quick rapid-fire questions and you can give me some quick responses. And I always like to start with the same one, which is as a company builder in this case, you co-founded Open Art. What would you say is one of the best parts about working at the company and what's one thing that you would like to change or are working to change?
I feel very happy that we are in this space that is that has new things every day. So when you work, you know, at Open Art, you feel like, wow, every day is a new day because there's new things going on. But to be honest, there's also an anxiety-inducing part about it because you constantly feel like you're falling behind. Um, but the the fact that you can like, you know, interact and then just, um, work on new things every day is very, very exciting for me.
Yeah, I I agree. And working at any AI startup right now must be so, so exhilarating because there's just so much changing so quickly. Um, what's one thing that you would change? I guess you did mention that it can be anxiety-inducing. Is there anything as a founder that you're working to change about the culture just to make it an even better work environment?
I wouldn't say it's I want to change, it's more we need to adapt because, um, we actually have been running Open Art as a very lean AI startup. You know, um, last year, for example, even early this year, we were like around 10 people. And so like our our revenue per employee is actually really high. However, I think we're at a stage that we actually need to grow like our team so that we can, you know, make sure we can like deliver what we want to achieve. So, that's one thing I'm trying to learn myself is like how you scale the team while not like losing the culture or losing the lean, like rapid, uh, iteration. Um, yeah. So, I think that's something we're adapting.
Yeah, that makes sense. And this is something I hear a lot of AI founders say is like there's a lot of things that AI can do now, but you still need people to build a great company, even if you need fewer of them.
Um, okay. I'm really excited to hear this next question. You must have seen thousands or or tens of thousands or hundreds of thousands of videos on Open Art. What is the most amazing image or video you've ever seen created on the platform?
Yeah, I love this question. Um, I actually, I would like to tell a story instead of a particular image or video.
Sure.
Um, so early this year, I was like, uh, actually just a few months ago, I was interviewing, you know, our users who use our one-click story feature. And then there's this, um, like, um, grandma from, like, New Zealand. And she's around 70 years old. And then she has like all the white hair. And then you can tell she's not as like technically advanced yet. You know, when I talked to her, she told me like she's been really, really enjoying Open Art. And then she started a YouTube channel of AI of of AI cats because of Open Art. She's been producing this like cat stories, AI cat stories. Um, while like, um, to to really like, not to make money, but then to give herself like, um, you know, like a creative outlet because she has retired and then she wants to be more creative. And then she started learning AI. And it was such a huge contrast to me that like, oh, this, um, really like nice grandma is running an AI cat channel on YouTube. You know, this type of contrast is just so, um, shocking to me. But like, I really enjoy talking to her. And then she is so open-minded to all of these tech, very different from a lot of people think, oh, like Gen Z must be like adopting these. A lot of these technologies. In my experience, a lot of these old people or like retired groups are really adopting AI to use as their creative outlet.
Yeah, I mean, in a way, there's a real advantage, right? Because they have more time and and so, um, right now we're all just trying to learn how to use these tools. And to have that amount of time and that, um, inspiration to want to create things is is really powerful. So, that's that's great to hear that. I mean, as a creator platform, obviously, your mission is to help people find that creative outlet and and bring beautiful things into the world, and it sounds like, uh, you've been able to do that. Um, if you could go back to day one of founding Open Art, what's one thing that you would do differently?
I don't know if it sounds greedy. I think we should have moved even faster. Like, I I think we are like, you know, fast learners and we move really fast, but, um, at this point, I really think being fast and iterative is probably [clears throat] like the the best advantage you can have as a Gen AI startup, especially in the AI application layer. So if anything, I would say we should probably like, you know, to move even faster and then, um, perhaps like when we do a lot of experiments, even like be more rigorous about learnings. Um, but I think in general, we're pretty good with like launching some MVP pretty fast. Um, but again, it's like how do you iterate after your MVP?
Yeah, I mean, given your error, our trajectory, you're certainly doing something right, but you can always move faster. And I'm sure your investors love hearing that because speed, speed matters.
Um, couple more here. How has your leadership style evolved as Open Art has scaled into a larger company?
I I think I like to share again a personal story, like, uh, about the leadership style because to be complete honest, I I like to be very hands-on. And, um, you know, it's a very good thing early on, but, you know, as you scale, like, it's actually really not scalable. I remember last year, like, you know, sometime last year, um, I was out of office, you know, for some personal reasons for a week. And I felt like it was so chaotic during the time I was away because like we didn't set up a system where like people could communicate effectively with each other. Like, because at that time, what we did was that we have daily sync. And then once people complete a task, they literally didn't know what's next. You know, we even didn't use or linear, any of these like task tracking systems. We just use daily sync as a way for people to sync with one another. It's it's honestly, it's pretty effective. But then at one point, you realize, oh, wow, but if if Google is out, out of office, then who will tell us what to do during the daily sync? You [laughter] know, it's going to be really, really messy. And then now we introduce more like task management and then also learn to be more hands-off so that people feel more autonomy. And honestly, my co-founder does a better job of giving people autonomy. So he constantly reminds me that, you know, to give people more autonomy and more freedom to experiment.
Yeah. We were at an event together last week and I and you use the analogy of a clockmaker and this idea that as a company gets larger, you can't individually manage every decision or every interaction. And so you have to create the conditions in which the company can thrive, much like building a clock or a watch and then handing it off to its user. So I really like that analogy and I think it makes a lot of sense. Last question, and this is a big question, but I'm going to challenge you to answer it in a minute or two. What role do you think creators will play in a future where AI may soon surpass human intelligence?
Wow, this is deep because I see the day coming.
Yeah, me too.
You know, like, um, related to this, right? Um, once I read a blog post which really like gave me goosebumps. Um, the claim is that if you think about like food nowadays, you know, you need to go to Whole Foods to buy organic food. If you just go to any of the grocery market, it's not organic food. It's like GMO food, right? It's genetically modified. And and then so the analogy is that in the future, perhaps like all the human-made like videos will be like organic food, and then majority of the videos you see will be AI-made. And I felt like, wow, like if you think about food or clothes, it's already happening. Why can't you imagine content be this way, you know? Like that day, I was like, really mind-blown by this analogy. And and ever since then, I was thinking, wow, so like the human-made content will become so precious in the future [laughter] because most of them will be AI-made, right? But again, like going back to your question, I I still think, you know, if you think about how of these, um, like AI models, uh, learn things, is that they still need to have training data. And and so I still think there need to be pioneers who are, you know, spearheading the content creation, like really create the original content. Um, but on the other hand, for the mass audience, you know, I think it's a good thing in the sense that they could express themselves with the AI as a tool. And so at that time, I feel like AI will become like, uh, like even though, you know, the the the, uh, AI could surpass human intelligence, I think humans have this like inherent emotional need that will still drive them to create. Like, you know, the grandma I was telling you, like she still needs the creative outlet, even AI could surpass her, like intelligence, she still needs the creative outlet. So I think to, like, in that aspect, then AI becomes, you know, a tool for human expression, like, and and just for creative outlets.
I couldn't agree more. I certainly think the outlet to create will continue to exist, but also I loved your analogy of Whole Foods versus the craft that you buy at a typical grocery store these days because there is a real difference. And I would like to believe that even once we reach Gen AI, you know, people will value stories and content created by other human beings. There's intrinsic worth in that.
Yeah.
Um, well, this has been such an amazing conversation as I knew it would be. Thank you for joining me, Coco. Any parting words you want to leave the audience with before we wrap up?
No, I I mean, I thank you for having me. And then I know like, Phil, you run this, um, business of helping people grow their subscription. So, uh, we'd love to work with you in the future if you consider us. And then I just want to promote your business to others. Um, yeah.
Oh, well, you don't you don't have to do that, but I appreciate you saying it. And I would absolutely love to work with you guys anytime. Uh, I'm I'm a huge admirer of what you've built at Open Art and would love to work with you any any day. Thanks again, Koko. This is this has been great.
Yeah, thank you.