# 300 Million Downloads (Matt Rouif, Photoroom) NEW ECONOMIES · 2025-08-16 Matt Rouif, co-founder of Photoroom, explains how the photo editing app grew from a two-week prototype in 2019 to 300 million downloads and $20 million in annual revenue on only $2 million raised, becoming the largest AI image editing tool globally. He details early tactics like localizing in 10 languages and conducting customer interviews at McDonald's. COVID accelerated adoption as sellers moved online, and Photoroom's focus on trust—avoiding AI hallucinations in product shots—set it apart for e-commerce. Rouif shares plans for a video editing beta and personalized advertising at scale, where infinite marketing assets are tailored per user. He identifies trends: specialized AI models for mobile speed and trust, AI video maturing in 2025-2026, and personalized advertising that creates unique imagery for each segment. ## Questions this episode answers ### How did Photoroom reach $20 million in annual recurring revenue after raising just $2 million? Matt Rouif explains that Photoroom achieved product-market fit in early 2020 by helping people sell online. When COVID hit in March 2020, everyone suddenly needed e-commerce images, and Photoroom's traction exploded. They had raised only $2 million but, fueled by this demand, quickly scaled to $20 million in annual recurring revenue within about two years, a growth rate that surprised even their Y Combinator partner. [10:29](https://neweconomies.podhood.com/75a99453-24e5-48a8-a6a5-4c910205f6e9?t=629000) ### What unconventional early growth tactic did Photoroom use to get its first users? Photoroom's founders, Matt Rouif and Elliot, took a hands-on approach by going to a local McDonald's in Paris during lunch. They would pay for people's meals in exchange for interviewing them about their photo editing habits and then having them test Photoroom. They also localized the app into 10 languages from day one and launched within two weeks, emphasizing speed and direct user feedback to iterate quickly. [4:50](https://neweconomies.podhood.com/75a99453-24e5-48a8-a6a5-4c910205f6e9?t=290000) ### How did Photoroom's technology enable the viral Barbie movie personalized posters campaign? In 2023, Photoroom's API was used by the Barbie movie to let users insert themselves into a Barbie poster, creating a viral campaign that reached 20 million people, including celebrities like Beyoncé and the Kardashians. Matt Rouif says this demonstrated how brands can generate infinite, customized marketing assets, a trend he believes will become the norm for personalized advertising, boosting conversion and engagement. [23:05](https://neweconomies.podhood.com/75a99453-24e5-48a8-a6a5-4c910205f6e9?t=1385000) ### Why does Photoroom prioritize trust and product accuracy in its AI image editing for e-commerce? Matt Rouif explains that for e-commerce, maintaining trust is critical, so Photoroom's AI is designed to enhance photos without altering product pixels or 'hallucinating' details. Unlike generalist models that might change a shirt's pattern or add non-existent food items, Photoroom ensures images accurately reflect the real product. This commitment has attracted major marketplaces like DoorDash and Deep Purple, who need reliable, truthful product representations. [17:25](https://neweconomies.podhood.com/75a99453-24e5-48a8-a6a5-4c910205f6e9?t=1045000) ## Key moments - **[0:00] Intro** - [1:06] Photoroom's origin: Matt Rouif failed with Photoshop after Apple's request for marketing assets. - [2:41] Photoroom reached 300 million downloads and became the #1 photo editing software globally in 6 years. - **[3:03] First Users** - [3:26] Photoroom went from idea to App Store in two weeks, localized in 10 languages, and added payments early for quality feedback. - [4:36] Photoroom founders interviewed users at a McDonald's and paid for their meals until security banned them. - **[5:40] Localizing** - [5:42] Matt Rouif localized Photoroom in 10 languages from day one after seeing Russia and China become top revenue markets for a previous app. - **[7:25] First Product** - [7:28] Photoroom's first product was a background remover that used AI to separate subjects from backgrounds. - [9:02] Photoroom's design principle: show, don't tell, to build trust by demonstrating value immediately. - **[10:00] Revenue Surge** - [10:05] With just $2M raised, Photoroom reached $20M in annual recurring revenue. - [11:05] Gary Vee promoted Photoroom in February 2020, then COVID lockdowns forced millions of sellers online, fueling explosive growth. - **[12:40] COVID Boost** - [12:44] Q: Would Photoroom have succeeded without COVID? A: It only accelerated the existing trend of e-commerce growth. - [14:20] Photoroom reached $20M ARR in two years. - [14:45] Photoroom later disclosed $50M in annual recurring revenue. - **[14:55] Positioning** - [15:16] Photoroom is positioned as the best image editor for online sellers, offering cross-platform native editing and batch processing. - [17:10] Photoroom's AI enhances product photos without hallucinating details, preserving trust for e-commerce sellers. - **[18:15] Global Reach** - [18:19] Photoroom's largest market is the US, with strong traction in mobile-first economies like Brazil, Mexico, and Southeast Asia. - **[20:00] Iconic Moments** - [20:14] After DALL·E 2, Photoroom pivoted in summer 2022 to generate realistic studio backgrounds, a pivotal moment for the company. - [22:18] Photoroom's next frontier is helping users create infinite, personalized marketing content, starting with its API for Barbie's viral poster campaign. - [23:21] Photoroom's API powered a viral Barbie poster campaign used by 20 million people, including Kardashians and Beyoncé. - **[24:45] AI Debate** - [24:45] Q: Will AI replace jobs? A: Matt Rouif believes AI levels the playing field for small businesses and enables scaling, not necessarily reducing headcount. - [26:46] Photoroom equips its team with AI skills by providing a budget for AI tools and fostering a culture of public experimentation. - **[30:00] Future Vision** - [30:00] Photoroom's goal is to go public and provide every business with an AI creative agency to boost sales and pride in their products. - **[32:30] Next Trends** - [33:18] Rise of specialized AI: Matt Rouif explains that fine-tuning models for specific tasks yields cheaper, faster, and more accurate results for mobile users. - [38:26] Q: If not Photoroom, what would you build? A: Matt Rouif would explore AI video tools. - [39:07] Prediction: AI video will be ubiquitous in 2025, and everyone will create AI videos by 2026, says Matt Rouif. - [42:18] Photoroom is beta testing video creation features, hinting at a possible 'Video Room' product. - [43:18] Personalized advertising trend: Photoroom enables infinite, customized product images, akin to a luxury white glove experience for every buyer. ## Speakers - **Ollie Forsyth** (host) - **Matt Rouif** (guest) ## Topics AI ## Mentioned Adobe (company), GoPro (company), Photoroom (company), Canva (product), ChatGPT (product), Claude AI (product), Cursor (product), DALL·E (product), Dusk (product), Figma (product), Notebook LLM (product), Notion (product), Photoshop (product), Replay (product), Slack (product), Stable Diffusion (product) ## Transcript ### Intro **Ollie Forsyth** [0:01] Okay, where can Photoroom go next? Can this get to 500 million users? Do you want to go public? This is crazy. And only 2019, so 6 years later, 300 million downloads, and I think now the number 1 photo editing software globally,right? Matt, welcome to the podcast show. Great to have you here. **Matt Rouif** [0:23] Thanks, Ollie. Uh, hi everyone. **Ollie Forsyth** [0:26] Um, well, hey, Photoroom, what a wild ride. $300 million now? That's insane. Um, finally, sorry, take us back. What's been happening and, uh, where are we today? **Matt Rouif** [0:38] Yeah, it's been a crazy ride, uh, indeed. I think the— well, if I look back at the Inception moment, um, Photoroom is not my first image and video startup. And what happened is I was, uh, in a startup that was acquired by GoPro in 2016, and we were editing the replay app. And really, for me, the founding moment, the Inception moment of Photoroom is, well, we were acquired, I was the PM for the video editing app, and I received, um, a little message, uh, a great message actually, email from Apple saying that we're going to feature the app I was managing at that time, uh, please provide some marketing assets by, by tomorrow. And I'm not kidding here. And, uh, well, basically we just got the acquisition, my designer was off, uh, and all the designers in California were sleeping. And so I, well, I tell myself, Matt, you're good with computers. Uh, I opened Photoshop and I tried to make the design I want. I had a clean story with the video editing app with GoPro, I think it was summertime, if I remember correctly. So, clear story. And the reality is I spent the afternoon, uh, struggling and, and trying to convey and tell the story I wanted. And it didn't make sense to me as like a PM, like both designer, a bit of development I do. Development got a lot easier, designer, designing didn't. Especially editing like these images. And I was telling myself, Matt, there's something to do here. We need to make that more accessible. We need to make, like, image editing more accessible to non, like, technical people, people not doing that every single day of their, of their work life. And, uh, that was like the Inception moment for Photoroom. Um, like a few months later, we started, uh, with Elliot, my co-founder, uh, with the idea of like, let's make image editing, uh, easier with AI. That was 2019. But really this idea of like, I want to create a photo, it should be easy with AI. You know, that's, that's the founding Inception moment. **Ollie Forsyth** [2:41] This is crazy. And only 2019, so 6 years later, 300 million downloads, and I think now the number 1 photo editing software globally,right? **Matt Rouif** [2:52] Yeah, yeah. Uh, if we look at AI image editing, uh, we're one of the top, like, uh, if not the top, uh, image editing software. Like, we have so many people from all countries in the world coming to edit their image with Photoroom. ### First Users **Ollie Forsyth** [3:04] And, you know, I can imagine a lot of founders who are listening, 300 million downloads, like, this is, you know, the dream,right? But every startup journey starts from somewhere. So take us back to 2019, you know, getting to those first 10,000 users, 100,000 users, first million users. What were those early tactics in acquiring customers at as little cost as possible? **Matt Rouif** [3:26] Yeah, I mean, so we started in 2019, and, well, what I think the first thing that, uh, looking back was, uh, really good for us is we were very fast to go from, like, idea, uh, to, to, to market. Uh, it took us, with Elliot, it took us two weeks, uh, to really from the first line of code to having an app on the app store. And when I say that, like, I mean, the product, I was ashamed of the product because kind of, uh, difficult to show to people, but we did put a few interesting things, uh, in the app. Uh, uh, we localized it in five, in 10 languages early, as something like it was not our first app. And we also put payment, uh, really early on, which was very interesting. Not because we wanted to make money, but because I think the, the feedback from people, like, paying, uh, is much more interesting than from people, like, just, uh, using your product for free. And so you want to listen to all of them. So I think the first thing is we were very fast to market to try to put something in the hands of users. Uh, what we did early is really spending time with the users. So our, our office was, uh, downtown, and we would go during lunch, uh, to the local McDonald's. And in Europe, it's more like, uh, you know, everyone goes there. So it's not like you can really find any customer, a lot of tourists too. And we just, like, ask, like, go in the line, ask people, what, what do you use, what photo apps do you use? What's your, like, what's your process if you want to edit an app, uh, a photo and put it on social media? Ask a bunch of questions, then, uh, ask them to test, uh, Photoroom, and then, uh, yeah, then collect this feedback and go back to programming. Talk to user, program. That was the early days. And we, we did it so much actually that the security guy from the McDonald's, like, prevented us, like, he said, you can't come. Okay, it's not possible. You can't. And we would pay the meal of the, sorry, we would pay the meal of the, of the, of the people. Like, okay, you're in line, we pay for the food, take what you want, and just want to interview you. So it wasn't bad for the business, but they, the guy didn't, he didn't want us to do that anymore. **Ollie Forsyth** [5:35] Yeah, they see you coming every day and you're like, no, go back. Go somewhere else. **Matt Rouif** [5:39] There are plenty of McDonald's in, in Paris, so we'll find, it would be fine. ### Localizing **Ollie Forsyth** [5:42] This is so cool. Wow. Okay, so unpacking this a little bit. So you localize, um, the product in multiple languages, like, basically from day one. Was there a reasoning behind this? You know, like, most founders are just trying to get their initial product off the ground,right? Um, was the main success in English or was it in a different language? **Matt Rouif** [6:02] Yeah, it's a great question. Uh, so I think the reason we localized, it's something I've always been, like, interested. Localization is something I like, and if you speak multiple, like, it's kind of what is your entire advantage? And speaking multiple languages, uh, it's one of them. Also, my experience from previous startup was that at some point, like, like, the US is clearly the biggest, like, market for app stores, uh, for apps. But at some point, uh, in a video editing app, I was managing as a head of product, uh, at some point, like, Russia and China were our biggest, like, uh, you know, revenue stream. **Ollie Forsyth** [6:39] Wow. **Matt Rouif** [6:39] For a different reason, but, like, don't, don't miss on that. And so, I mean, I think that in, in the past, like, it was very, like, I already had a process, some script, uh, uh, in mind, and so that was quite cheap for me, uh, and for us with Elliot to do it. That's the first step. Already wrote like a bunch of blog posts on how to do it. And I think for Photo App, it's kind of knowing your industry, like, it's very visual. So, you know, the difficulty, like, people look for the visual first. The difficulty of translating the app is not as much as if you're, like, let's say you're a calm or you're a blinkist and you have a dual lingo where you have ton of content to write to explain. So I think it's depending on your industry, knowing what your user wants and seeing what, what's easy and what's, what can be an entire advantage to do. ### First Product **Ollie Forsyth** [7:28] And so the first product,right, was this, um, background remover. So basically, like, for listeners listening if they didn't know, it's like, if you have a blue background and you want a blue background removed, you go to Photo Room and they will remove the, the background,right? **Matt Rouif** [7:41] Yeah, yeah. Let's, I mean, we didn't talk about that. So, yeah, we really started with, like, a simple product in two weeks, but we, like, Photo Room today is the best app in the world for AI image editing. And in particular, if you look at algorithm, we're the best in the world if you want to remove product, take a product shot, uh, put it on a white background, create a shadow, anything you do in a studio. And the first version of the app was really just one view where you would be able to split the background from the, for the subject, uh, uh, like, apply AI to that. And it happens that in 2019, kind of the state of the art was getting really good at it. And so we, we, we started with that. And there clearly, there was an audience as we got better at this algorithm. But the app was just one page, you know, one view. **Ollie Forsyth** [8:21] Yeah. I, I tried the product before I came on, and what I love is how easy and simple it is to use the product without even having to sign up. When you use different products and similar products, um, you're like, oh, okay, it's completely free to use, but then to download the image, you have to sign up. There's like a cash. And what I love, I think, about one of your product principles is everything should be done in like a couple of clicks,right? And you can really see that from even just trying the product within a few seconds. **Matt Rouif** [8:48] Yeah. Uh, well, in visual editing, that's one of my learning from past, uh, you know, and from Photo Room is like, people know it when they see it. So you really want to, like, show, don't tell is another principle you could use, but you really want people to see what's happening. And that's how you build trust. And really, already in 2019, there were so many apps that were like advertising something, like this kind of feature, but just not delivering on it. And we told ourselves, like, we just want to show the value to the user to build trust. So they, they come in the product and they trust, like we, they come from, like, yeah, good karma, you know, they see like they're, they're, they're not lying. So if they're trying to do something, uh, if it doesn't work, it's maybe it's a bug. It's not like they're trying to, uh, rip me off. And so that we really wanted to come from that. And yeah, and the idea is we are going from not for non-designer. Remember the Inception story. So people, they don't feel they're, they don't feel confident in their craft or non-craft of designing image editing. And the goal for us is to get the user as fast as possible to a situation of confidence that actually they can do beautiful stuff. And so the idea is you need three, four taps in the app to really have this wow moment where, okay, actually I can do some beautiful stuff. ### Revenue Surge **Matt Rouif** [10:03] I can do it by myself. **Ollie Forsyth** [10:05] Love it. Well, youths are loving it. And, um, VCs I think are also loving it,right? Um, you raised $62 million to date, give or take, and the sound I found, found quite interesting. The first $2 million you raised in venture capital, you got to $20 million in annual current revenue. Normally it's the opposite way around. Normally you spend $20 million to get to $2 million in revenue. Wow. What happens? **Matt Rouif** [10:29] Yeah. Yeah. It's, uh, I, I think it's actually you see AI companies doing, uh, a bit more than that, but I remember talking with our Y Combinator, uh, uh, partner saying like, I mean, it's the first time we see a company growing that fast with that little of capital raised. And really, like, in a few months. So like back to the story, 2020, uh, we start to be like, we launched in 2019, like a bunch of iteration, get the, the product to work better, get the AI, we trained our algorithm to be better. And early 2020, we really start to, like, have some traction. I remember Gary Vee talking about Photo Room in, uh, February 2020. Really like, if you're a reseller, if you're a hustler, that's the app you need to use. And we had some other YouTube influencer. And so we really got traction in February 2020, like product market fit on people selling online. And they need good photos to sell. That's, that's the learning. I'm starting from like, I'm a product manager, I need an image editing software. And like, it's in 2020, it's like, actually, it's not only a product manager, it's like a lot of people, like hundreds of millions of people. Uh, you know, eBay alone is 20 million sellers. And, and so we have that. And, well, that's kind of, well, you need some kind of luck. So at that moment for our startup to be grow that fast. And what happened is March 2020, uh, well, COVID hit. And, well, it's been painful, I think, for a lot of people here. Uh, not easy to deal with, like with a kid at home on my side. Uh, but for Photo Room, like at once, everyone was selling, like in, in the real world need to like put their product photos, their commerce, their e-commerce. Well, they need to build the e-commerce and have some like a web presence to sell what they're building. And so everyone at once need like, uh, imagery, uh, to sell. And we had like people that couldn't meet, uh, the end of the month and they'd tell us like, you know, thanks to you guys, uh, like, I can pay my rent for my shop and, uh, and for my apartment because I was able to put my physical store online. So we really got a lot of traction from that point. And so that's, that's the first thing that happened during YC and just after COVID started. ### COVID Boost **Ollie Forsyth** [12:44] This is like a, um, COVID success story,right? I mean, there are a couple others. Um, Hopin was another one for like events. I know it's not here anymore, but, um, there, there are a couple of like success stories,right? If you, if COVID didn't happen, do you think you would still be at the stage where you are today? **Matt Rouif** [13:04] You know, it's a good question. Um, I think it wouldn't have gone as fast, but I think like the need was here. Like, if you look, well, so, so a big part of our users are people selling anything online, creators. Like the trend is positive. Like, uh, you know, and if you look at the trends of like, uh, e-commerce compared to like global, uh, commerce, it's kind of a step function for that. So at once everyone like asking it like themselves, like, do we, uh, what, how do I change? And so it kind of opened the mind of people, but it's only acceleration. Like if you look back and look at the 10 years or five years, uh, timeline, then I think it's similar. It's only accelerating, but then it kind of catch up, like, you know, like a Zoom had some difficulties after because it kind of, uh, went back to normal level, like higher because it's a natural trend. But looking back, it seems linear, like in adoption of, uh, online, uh, e-commerce. So I think it accelerated us, uh, clearly. It also brought more competition. Um, we're better off. I don't think it changes like a Photo Room history on the 10 years, uh, you know, but there are definitely ways for startup that you need to happen to, to be, I mean, we're lucky to be at theright time,right moment. **Ollie Forsyth** [14:20] For sure. It's insane. Yeah. It's insane. I mean, even just to get to the $20 million in annual current revenue, I mean, how long did that take you? **Matt Rouif** [14:31] Yeah, it took, um, so we raised, um, it's 2022, if I'm, uh, correct, a series A, and it's a bit before. So I think it's like, it took us two years. Yeah. **Ollie Forsyth** [14:45] Wow. That's crazy. And the latest is 50? **Matt Rouif** [14:51] Yeah. One year ago, we talked about 50. **Ollie Forsyth** [14:53] Yeah. Wow. In six years. But when I think about like these photo editing tools, it's like, okay, there's Adobe, there's Figma, there's Canva. They all have different use cases,right? How are you describing Photo Room to your audience? And where do you kind of fit in this whole, you know, um, creator tech stack? ### Positioning **Matt Rouif** [15:16] Yeah, that's a great question. Um, so Photo Room is the best image editor if you're selling online. If you have an e-commerce, if you have a physical store, we're, we're really the best tool, uh, for a few reasons. First, we are like cross-platform. We don't have that many, uh, native. So you don't have like Photo Room is the only native cross-platform app that you can use. And we have a lot of users when they sell online, they take photo from their phone and live, you can have under your web app or your web photo app, like the image. So you can like really download it, put it on eBay, Shopify, make some changes, add some text. And we really have this full flow where you can start taking photo from mobile. You can explore on social media from, like, stay on your phone, but you can, you can really go all the way to the web. Uh, what we, a few other things that we do are like we handle teams, you can collaborate. We have a batch editing tool, like, you know, when you're a commerce, like the image are your best proxy for what you do. So it's about like consistency. Like if you look at the window, uh, of a shop, like it's how is it, is it tidy? Is, is it clean? Well, it's the same for online. Like you need good photos. And, and this is so important. Like it's really the best proxy to the craft and the attention that is behind, like the person selling. Uh, we have batch, so you can under, like, you can edit hundreds of photos at once. So there are a few specificities. And the last one, maybe the AI is really, if you are thinking about commerce, e-commerce and marketplaces, like it's about trust. Like a lot of things about e-commerce is about like, do I trust the person that is behind that? And like over time, we put a lot of, well, first is like it's someone you know, like local commerce, you have trust, but then you build like this kind of proxies for trust online, which is like, you know, ratings, uh, good imagery. And, and Photo Room, like Gen AI has been a lot about, well, let's be creative. Let's, let's, let's create something that the world that doesn't exist. Can I trust that? And Photo Room AI has been a lot about like, okay, let's enhance the photo, but not lie on it, not hallucinate at all. So if you have a, if you have a product photo, we won't change the pixels of the product, because otherwise you get like, you break the trust of the user. They receive like something that the pattern that is different, uh, uh, with a like, like a drawing on the shirt or food that has peanuts in the plate, but doesn't have them. So, you know, that's why most, like a big, that's something recently in Photo Room, but we also sell on marketplaces for B2B, but the really the, uh, DoorDash, the Mercury, uh, the Deep Purple of the world, some of the biggest marketplaces, they use Photo Room because you can really trust the photo and it doesn't like put all photos like to be wonderful, uh, as you like Gen AI would usually do. Just it's reflection of the craft and the initial photo and the initial dish quality. So there's this like, okay, how do we make the photo to be the best representation and build the trust with the user, which is very different and very important for e-commerce. ### Global Reach **Ollie Forsyth** [18:19] Um, first of all, like, it's just crazy how fast you've grown. It's, you know, there are so many more people now becoming creators, um, who need, you know, products like you guys. Um, where would you say your biggest markets are today? Is it the US? Is it Europe? Is it West, you know, is it Eastern markets? How do you stand out in kind of the field? **Matt Rouif** [18:41] Yeah, the, I mean, the US is for all mobile or SaaS company is clearly the biggest market always. So US is the biggest market for us. I think what sets Photo Room apart is we're very popular in mobile first, uh, economies and yeah, places where people have a lot of, uh, you know, side kick, uh, side job and like, uh, have something like to make a bit more money at the end of the month. I mean, just to give you an example, I think it's, we published a report on, uh, like the Gen Z, uh, being financially independent with, uh, with like, uh, your e-commerce, your, your small businesses. And in the US, it's 50, almost 50% of the Gen Z that have something on the side where they're trying to make money. Um, so young, uh, population in the US, but also in Brazil, in Mexico, in Southeast Asia, which are, or even Japan, which are big countries from like mobile, uh, these are big countries for us relatively to other, because you have like this full, yeah, generate, people have three jobs and like, they create a lot of businesses. Like there's a lot of, uh, unofficial, like, uh, uh, jobs that are and businesses that are started and they run their business from their phone. So we're very popular in, in this and these are big regions for us. **Ollie Forsyth** [19:59] This is amazing. Um, when I think about these types of companies, there's always been almost like an iconic moment or like there's always been like a biggest moment where it's really like taken the company from zero to one incredibly quickly. ### Iconic Moments **Ollie Forsyth** [20:14] If we take away the COVID story, which one moment do you think has really accelerated the Photo Room story to where it is today? **Matt Rouif** [20:24] I think the first one is indeed like a COVID and background removal. And now there was a first one where we like really create, started to create the second one where we really started to create like a full generation, uh, image editing software, which is, um, summer, uh, 2022. Um, I remember there was DALL·E 2 that was out a few months back and we were like planning with all your, like a founder on like what's next. And we did a deck where it's like really it would be amazing if not only we can have a white background, but we can also create like what you do in a studio, like a full photo, like, uh, you know, a studio shot costs 10,000, like a cheap one costs $10,000. And, and really putting down like during the summer, putting down, okay, this is what's next for us, like really generating the background. Uh, and a few weeks later, there is stable diffusion that, uh, open sourced, uh, their, their generation model after DALL·E. And it was quite good. And I remember like literally posing the company like, you know, we need, that's, that's what's next for Photo Room. Actually, that's, I remember we said, it's probably like one of the biggest like moment of, uh, the photo industry for at least a year, but maybe the next five years. And, and so we just like, okay, let's do that. Let's work on that. Let's make it accessible to our users. And it took us, well, the first version took us three, three months, but then it took really, it took us really six months to get the best product for our user out there. And that was like the, like the second, like stopping the company for that and really focusing on that was one of the best and most important, uh, moment of, uh, Photo Room afterwards. **Ollie Forsyth** [22:01] It's really cool. And you know, today you, you offer so many products,right? And, um, what comes next after kind of the photo editing? Is it video editing next? Like, you know, what types of products do you think is missing for, uh, creatives, creators, people who are running e-commerce companies? Because you guys have now got that distribution,right? Which is what everyone wants. And now it's, you know, you have so many, you have so many new opportunities that you can tap into. **Matt Rouif** [22:28] Yeah. Um, I think what's next is really about, like we really are the best for, uh, this product catalogs photos. And the next part is like, how do we help people to create like, uh, marketing content? I really believe in this idea that, uh, you can create infinite contents if you're running a marketing campaign, like really adapt your imagery, your visual to who's seeing it. And the iconic moment for us for that, uh, is our API. So before, after the apps, we started an API that does the same thing. And it really got traction in 2023 with, uh, the Barbie movie. And so basically, well, what they did, they used Photo Room technology to put everyone in a Barbie poster, which is pretty cool. Uh, but at the same time, you can scale. So 20 million people use that. Uh, the Kardashian, uh, Beyonce, it was crazy. And so that's like, that's a very, very viral moment. But when you think about it, what you see is you see the marketing of a product with your, like, with something that is customized, is your friends or people you know that are in the poster. And I think this is the future. Like you have infinite, so 20 million in this case, ads, like marketing assets for a product and everyone sees a different ad. And we are the first to do that in 2023 with Barbie, with our technology. And I think that will, is going to become like accessible to everyone. Like how do you do marketing and specialize your marketing asset to every ad, every person, infinite marketing assets. Um, an example is if you're a, uh, we work with one of the big, uh, furniture marketplaces. And this is like, how do I, like, if only you're looking for a chair you want to buy, how do you see that in something that looks like your living room automatically? And how can we make for each segment a different photo? And we start to see people and businesses like really customize, combine, create like infinite variation of what they're selling. And this is working quite well. It increased the GMV and, and really that's the future of what we build for e-commerce and, and creators. ### AI Debate **Ollie Forsyth** [24:45] That's amazing. Um, I can see this working amazingly well also for creators as the kind of, the whole marketing tech stack starts to change,right? People want to spend marketing with people who they trust. Um, with the world we're going towards, with AI replacement or enablement, where do you think the space goes next? Are some jobs that are going to be replaced, but a lot of people are going to use AI to obviously increase their productivity and enablement. Do you think we're going to see a mix of the both or where, where do you stand on this? **Matt Rouif** [25:22] Yeah. Um, so we're making cheaper, like the creation of high quality, like, uh, almost common use cases. People didn't have money, uh, to go into a studio shot or I'm a marketplace and I can't pay for the studio shot of, of all of my sellers. If you're DoorDash, you can't pay for that. Soright now it's things that were not possible before that are happening. I think AI gives some controls, uh, for bigger companies to be able to scale their process. So I can all, like being transparent, I think you empower the small businesses to do more and we'll have more and more small businesses. You're leveling the playing field. You, uh, enable the bigger company to scale. So they really need to think about, okay, how do I, uh, do 10X more and probably with a bit less people, I think in some of the, the processes. Um, but I think it's still employ the people. Like, uh, the way I think about it at Photo Room is we have like a four people team doing hundreds of like assets for us, uh, marketing ads, uh, for us and, and for support, for instance, we have AI doing part of our support, but at the same time, like we want to employ, like we didn't get a smaller, we didn't, uh, decrease the size of the support team. We're actually trying to have them like empower even more our, our users in support, telling them, oh, you know, you could use that. So like going to more advanced tasks, uh, I think is the way it's going to go. **Ollie Forsyth** [26:46] Yeah, it's pretty interesting. We actually spoke to, uh, Wade Fossa, the co-founder of Zapier, um, in, uh, a couple of previous episodes. Yeah, exactly. And, um, you know, we were talking about, okay, how the, the future of, um, hires inside companies, do they, how well equipped do they need to be with AI? And jumping onto, you know, your, your team culture now, um, how are you, are you guys doing much to equip your team with the latest AI skills? As what Photo Room has done is not only incredible, but it also empowers so many more creators,right? Creators, creatives, and people with e-commerce companies. But with that, I imagine your team need to understand the latest of what's happening within AI. So the question is, um, how are you guys equipping your team with the latest AI skills? **Matt Rouif** [27:39] Yeah. I, I mean, if you want to invent the future, I think the best thing is to live in the future. So, uh, like give the tools. We really think about if you need any AI tool, like please buy, like we give everyone like, of course, a credit card to build any, any AI tool they use. And that's, that's kind of the best way for us to like project ourself in what's going to happen. And so we can build what are the pain points for the future. And, uh, and actually there is a good, uh, podcast of one of our, uh, of, uh, the lead career of our user research team. And she mentioned how much she, like, before joining Photo Room, like, uh, people were telling us, you need to use AI. And like, our boss was like, you know, your goal is to use AI. And she never really managed to have that working. And then at Photo Room, she was like, like she, she was emulated with like other people, like doing exactly using AI and seeing how people get creative with like new, use, new tools that you can build. And she really redefined the way we do user research. Like we have an internal podcast that is, um, driven with like an AI jingle that tells you about the, how users are using Photo Room. Uh, we use Notebook LLM to have like, uh, instead of reading a Notion document or a PDF of the result of the user research, your Notebook LLM is going to tell you like, uh, like generating with AI to people talking about the results. So it's much more engaging for people to listen to it. And we have many tools like, uh, Dusk. We of course use ChatGPT, uh, coding with, uh, Cursor or Claude AI. So really encouraging everyone and being very transparent about the way you use it. So we, to give you an example, we love using Dusk also, which is an AI assistant for your company that plugs itself into your knowledge, uh, on Slack and Notion, uh, because we can do it publicly. And so you have a channel where people are asking like, uh, okay, Dusk, can you give me the answer for that? And it will give you the answer, but yeah, that, what's, well, first it answers your question, but second, you also see, you also see people, uh, asking questions to the AI. So it's encouraging for newcomers to see, oh, okay, that's how we can use AI. Because people, okay, uh, you know, sometimes it's difficult, it's change management. It's, it's difficult to see how I can use AI, but seeing other people doing it is the best way, like social, uh, proof and social, uh, patterns and seeing others doing it is the best way to encourage people to do it. **Ollie Forsyth** [30:00] Totally. I think, um, you know, just seeing how quickly this space has changed,right? In just the, the last few years, it's, it's hard to keep up with what happens next,right? We saw generative AI in 2022 when, um, ChatGPT came out. ### Future Vision **Ollie Forsyth** [30:18] Then, uh, last year was kind of like AI agents. Now it's vibe coding. Um, where does Photo Room go next? So take away like, um, you know, the, the marketing campaigns, which we'll talk about next in the trends. You know, where can Photo Room go next? Can this get to 500 million users? Do you want to go public? What's the big goal? **Matt Rouif** [30:41] Yeah, I mean, the big goal is to go, uh, public. Like, uh, I was already part of the company being acquired, uh, we're very ambitious and I think there is space for AI native companies in all industries. So we're very ambitious on that. The, the vision, uh, I mean, the vision we have is really for any, for every business, you have a creative agency in your pocket, uh, on your laptop that is, and even when you sleep, like it's working for you and helping you get the best out of what you're building. And it's really important, like when you see the eyes of, I mean, it's not only about like selling. So you help people sell more. We have restaurants selling 20, 30 percent more because they use Photo Room for the imagery. But it's also the look in the eyes when they see like a good representation of what they're doing. It's like you have a creative agency that is, it's like you go to a studio to get a good picture of yourself. You look more beautiful,right? But it's the same for the business. It's your baby and it looks beautiful because you use Photo Room and it's incredible how changing it is for them. So we really want to push that idea, uh, forward and empower these hundreds of millions of business owners to, to be proud of what they do. **Ollie Forsyth** [31:48] It's insane. And you know, what you guys have done in just what, five, six years, um, I'm really excited. I think, you know, with the creative economy now, there are 200 million creators globally. Everyone wants to become a creator. They need the best tools on the market. Um, and I think a lot of this is just how do tools make it so easy for them to become successful, um, in the cheapest possible way. And what I love about Photo Room is from the pricing perspective, like it's affordable for everyone. You know, when, um, uh, from like the VC days, it's like, oh, you're, you know, looking after SMBs. There's no money in SMBs, but look at you guys, look at HubSpot, look at Zapier, and there are probably a bunch of others. Um, amazing what you're doing. ### Next Trends **Matt Rouif** [32:31] Thanks. **Ollie Forsyth** [32:32] Let's see what happens. Um, cool. Well, let's move on to the second part of the episode, which is talking about the next big trends in tech. And I think this is the most fun part of the episode where, you know, we'd love to, uh, understand what trends are, are our guests thinking about and the types of businesses which could be created from those trends. So we've got a couple of trends which we're going to talk about today. Um, one is around the rise of specialized AI. Super excited to chat about that. The second is improvement on AI video. And the third, um, personalized advertising at scale, which is kind of what we just spoke about. So let's start with the, uh, the first one, the rise of specialized AI versus generic models. What are we thinking here? What's your take? **Matt Rouif** [33:18] Yeah. Um, it's very interesting. Like you get super powerful models that can are very generalist and Photo Room like always thrive on making, with making like people like very specialized, you know, uh, to the user need. And even early on, I remember our, uh, white coordinator partner telling us like, you know, Photo Room is different because most of the labs they're building like they're pushing a new technology. Photo Room is different because you are building useful AI, which is a bit different. And well, what we do is we look at what generalist, what generalist model can do, but then we always train like when we get to the limit of that, we're, okay, let's, let's make it deep versus horizontal where it's needed for users. And that's where, why we are the best in the world in background removal. And when you see, you know, there are a lot of like new tools that are, uh, for instance, generating text to image. Well, it's still weak in some dimension. You can see how, well, if you want to like respect text or be photorealist or respect, uh, be good in, uh, I don't know, French language or Asian languages, it's not really good. And most of the models, not only photo, they can get very, they can get much faster, much, uh, cheaper. If faster is also cheaper and also most, uh, perform a bit more if you specialize them on, uh, one need. And so this is really something that has been good for us is train like be a bit different in the way like we do AI backgrounds for studio shots where we really specialize our AI background models to do what is like not hallucinating the product, not creating new parts of the product, but be very realist and realistic in the, in, in the background. And so it's the same with text. Like you want a model that is able to like, uh, write like, uh, let's say recommendation of images for code. If you want an image, uh, model to be able to write the diff and really make the changes from the code that you need and being good at programming. So you see like there's like the generalist model are very, very good, but then when you apply them and, and, uh, make them like a, you know, small fine tune, make the compress the models on wine dimension. That is very important. You get cheaper, better, uh, faster result for the user. And especially when you're doing mobile, like having something specialized is, is key. So I think there'll still be this duality, but in the past year we saw like, okay, it's going to be like a lot of, uh, the talking has been about the generalist. It's good to discover new use cases, but then when you want to get really applied, you want to cut cost, you want to make it fast for mobile usage, then the specialized are very good. And for commerce, as I told you, trust is very important. Not changing the, the, the details of the product are very important. You have a OpenAI release, the high fidelity, you know, image editing. It's not really high fidelity, like compared to the way we train the model, it's very like, it still changes the product. So there's a lot of, uh, potential in like really getting your model really good at something and making it faster and smaller, the AI model for them. **Ollie Forsyth** [36:31] Yeah, it's something I think about like the accuracy,right? Uh, I, I, I used a, um, a well-known image editing tool, should we say, the other day. And, uh, it wasn't background removal, but I wanted one, one side of the picture on the other side,right? And with the prompt, it's still not 100% accurate. Okay. We, we have moved so fast in the last few years, but how far do you think we are away from being almost 100% accurate across all these models? **Matt Rouif** [37:00] Yeah, I, that's, that's a great question. Um, it's like the, the progress that the models are making is, uh, crazy in a few months. And like if you look back six months ago, like people wouldn't have to assume like it's impossible to do that. I, I think we're still like, uh, quite far a few years to like really have, uh, full control of like if you take design, the spacing between, uh, for instance, uh, letters according to the designer would use, I think it's still very difficult to say, okay, I'm exactly going to control that. Or if you think about marketing, like, okay, text generation is going to be good for like one or two sentences. Uh, if you want to have a full paragraph and it's legally binding, I mean, there's some noise on that. It's diffusion. By definition, there is like some, uh, probabilistic part of rendering. And so you, it's not sure you'll get like on 1,001 million images, you get itright all the time. So there's still be some the heuristic or deterministic rendering of images. And it's one of the challenge of AI companies is making sure you, you make a mix of the two when it's, it's important. **Ollie Forsyth** [38:10] For sure. It's, um, it's so early days. And by the way, the tool I use, it wasn't Photo Room. Don't worry. Your tool is, uh, awesome. Um, if you, if you weren't building Photo Room today, what type of product would you be building in this kind of specialized AI space? **Matt Rouif** [38:26] Oh, that's a good question. Um, yeah, I mean, you know, uh, we didn't talk too much about it, but of course when GoPro, the company I was at, we were doing video. So I think video is really hot and relates to the next one of your questions, but, uh, I think I would explore like what can we do with video that is, uh, interesting, uh, if I would start from scratch, uh, today. **Ollie Forsyth** [38:49] Cool. Let's jump into it. So improvement on AI video. I think this space is going to change a lot. Like we've seen a lot in, uh, image,right? I mean, you guys know it as well. Video is surely next. Um, yeah, improvement on AI video. Take us through what, what are you thinking and what's yet to be achieved? **Matt Rouif** [39:07] Yeah, I think 2025 is really the moment where you can see like good videos, uh, everywhere. Uh, I don't think everyone is still doing these videos. So maybe it's like, like seeing AI videos everywhere is 2025 and, uh, and seeing like everyone making AI videos might be more 2026. Um, and well, what's the, the challenge is, I mean, the models are really getting really good and realistic. So you, you have like trends, uh, are coming from that. It really gets you to animate and put in motion everything that exists. So it's something we, some of our users are, are playing with in Photo Room. Like how do I animate that? I, I think, yeah, basically the, the technology like VEO3 runway, they're really good at that and you'll have more tools. So it's coming, uh, and it's good enough to like, well, not see the difference for someone that is not spending a lot of time from, uh, normal content. What's, what's crazy is also creators making like stories and telling stories. The same way before you had to be a, like before you had to write something and the only way to create some really good, uh, fiction would be the text medium. And it seems now you can be alone and do like a 10, 30 minutes movie with CGI with the, well, yeah, with, uh, with a good prompting and it's going to unlock so much creativity, so many stories that were not possible to tell before, imagine. So I'm very excited about that part. Um, it's still like video is, um, it's like three-dimensional, like, you know, a photo plus time. And so it is like one of the challenge we add at GoPro is teaching people to tell a good video story. And so as for photo, you want to show kind of, people know it when they see it. So you kind of want to show what, what would be possible as a good video to your, uh, users and then they decide. And so this is very expensive, uh, still. So that, that I think is one of the challenge is like, uh, it's difficult to show the user what it could be because like, uh, it's so expensive in GPUs to, to render a video. So that's, that's I think the biggest limit. Uh, cost will go down, you know, five, 10X win in one, 12 months is, especially for new algorithm is common. So, uh, that's why also I think 2026, 2027 are more the time where we'll see everyone starting to create on their phone videos. Uh, today it's difficult to run like a different, different possibilities. Um, and I think video will serve like an authentic story that people want to tell. Like, uh, I really believe like people are always in for a good story. That's the common part. They read the story of Disney, the story of Pixar. It's always the part. So I'm very excited about this, but, uh, it's still a bit expensive, but I think we get to the level that is good for video creation in this, this year. **Ollie Forsyth** [42:13] So we have Photo Room. Are we going to have Video Room soon? **Matt Rouif** [42:18] Actually, uh, we're, we, yeah, we're beta testing. So some people will be able to experiment with video creation into, into the appright now. **Ollie Forsyth** [42:27] Cool. One, uh, part of video content, which I think has been amazing is creators can now truly be global from day one. So matters to creator, if he only speaks English or French, he can now translate his language into his content into multiple languages. Right. Kind of annoying. I had the idea like four, five years ago, but I never did anything with it. Uh, but you've got like the likes of 11 Labs now, uh, who are doing super well. And this just, the whole space is changing so quickly,right? And it kind of, uh, jumps onto the, the third trend, which we're both excited about, which is around this like personalized advertising at scale. Advertising feels like it's coming back. People are spending more money with mostly creators. Um, but it feels now like with AI, there is so much like insane opportunity. So let's dive in. So what is the trend? And if you were to start a company in the space, what would it be? **Matt Rouif** [43:18] Yeah, I mean, I think it's like infinite content creation. It's really what is coming. And so the same way, well, for video, what is interesting is you can tell the story of the, of your, uh, I mean, with your kids, like I print books, you know, with my kids telling them, sorry, I wasn't able to do that before. And like really, it's, it's kind of crazy how you can do that. I think for ads, you get to create unique imagery, uh, for, yeah, for your users. Uh, it's kind of the same experience that you get for luxury brands that are really white glove, where you get like a dedicated service. I think we can do the same thing for like the way you, your touch point with your user are going to be. And it's really about creating a lot of content and adding it, well, first to all the segments that you have. So I mentioned if you're a furniture seller, you want to have like different scenes, like if you, I don't know, if you like minimalistic design for your home, if you like, uh, Scandinavian, if you like a more cozy place or, uh, countryside, like there are pieces of furniture that could work, uh, like in all these styles, but I want to see it in what matches what I want. And, well, the first trend is, okay, depending on where you are, we can serve you like, uh, good suggestions and that gets you like a higher conversion rate. And then the future, I think, is really about, well, you know how all the emails that we send in your, uh, CRM are personalized, like, "Hi, Ollie, are you doing today? You left this item in your basket." Well, you know, Photo Room API is empowering, uh, you to get an email where this is actually an image and you see the product. You might have like another product, uh, side by side in the photo and we tell you like a big 10% discount and you see how they go well together and with some kind of context that we know speak to you. And so I, I think this is going to like become really empowering. It feels like very customized, really personalized, uh, to, to, to the seller, uh, to the seller and the buyer. And, and this is like, uh, really having this unique representation of both sides of the transaction is very interesting to me. And we already see like the most successful companies today in e-commerce or even apps marketing, they're doing hundreds, you know, of assets every month. That's how you, like you feed the meta Google algorithm with hundreds of ideas that you get. And then, well, the algorithm would pick what's working the best. But if you want that to work, you need to like create this volume of assets. And the way to create that is to use AI companies. Like that's the only way you can both create content that is, uh, customized, but also respect your brand. The same way you say for local, like, uh, you know, uh, translating in 50 languages that increase the conversion for, but then it's like that's specializing yourself to your 50 segment, but then you can have segments that are not only language, like who are you, what age, where do you live? And that, that is so powerful to build that. **Ollie Forsyth** [46:22] It feels like there's a massive opportunity. Like there are, the more I think about it, there are quite a few like AI agents for marketing, but no one's done like the full spectrum yet, the full tech stack within the marketing. Uh, that could be massively impactful, especially for creators, SMBs, bigger companies, um, connected to your CRM, given understanding who your audience is, automatically, you know, create visuals and your marketing campaigns could be quite, could be quite neat. **Matt Rouif** [46:47] Yeah, I think it's going to be incredibly powerful. And already we see it, like the most successful companies are the ones that are creating tons of content. So it's already happening. Uh, we'll get smarter over time and more, yeah, more specialized, more, uh, more quality, more content, more video. And that's, but basically your marketing agency, your creative agency, uh, in, in your pocket working for you while you're cooking if you're a restaurant. Like, uh, I mean, if you're a chef, you don't start a restaurant to be building creative assets. You, you want to cook,right? **Ollie Forsyth** [47:15] Exactly. Well, hey, we've covered so much. We've covered the founding story of Photo Room. Amazing to see what you guys have done. Uh, the trends we're really excited about and the types of businesses which could be created from those trends. Um, super excited for you guys. Um, you know, let's get to the 500 million downloads IPO at some point. Um, for any of our listeners who want to find out more about you, where, where can they find more? And, uh, they should definitely check out Photo Room, of course. **Matt Rouif** [47:40] Yeah, yeah. Uh, please first download or use Photo Room on the web. Uh, if you're, uh, marketplace or large e-commerce, we can also definitely assist you with our API. And, uh, I'm mostly on X and LinkedIn, uh, mature, uh, with, uh, and yeah, happy to chat. Like, uh, I love to talk to our users or also creators. **Ollie Forsyth** [48:00] Awesome. Matt, thank you so much for your time. **Matt Rouif** [48:02] Thanks, Ollie. --- This library is powered by PodHood (https://podhood.com), the podcast website platform.