Podcast

Dhaval:
This former Dropbox engineer built an AI product to create stylized professional product photos for people running e-commerce shops. Boyang is a founder of Stylized ai. He Figured out a way to leverage depth extraction, AI and 3D rendering to empower e-commerce sellers to transform their phone photos into professional assets for everyday use. One of the biggest lessons I gleaned from this conversation is how he approaches product development by focusing on areas of growth and applying the user mindset to product creation.

Hey, Boyang , talk to us. Tell us about your journey. How did you come to, identify this, space? And, tell us a little bit about you first and then we can talk about the product.

Boyang:
Yeah, absolutely. Great to be here. Dhaval I joined Hive. So Hive was a social media company back in 2015, building like a Twitter clone. We pivoted hard to enterprise computer vision SaaS. And so that’s where I first like learned about ImageNet, um, did all this AI stuff built like a trading pipeline and all of this stuff. And then after that, worked in some productivity tools at Dropbox, worked at e-commerce at Square, and then for the latest venture, I like put all those things together. And really it’s about targeting the market that I care about, which is e-commerce sellers, which is a huge market, right? You get a lot of customer iterations, a lot of customers to go after. You don’t have to really be scared to approach any particular one. And that’s a great feeling when you’re just getting off the ground, right? Because you need that iteration cycle. And so we, we. Sort of putted around me and my co-founder, looking for ideas that really resonated. And one of the things was like, oh, people want to take photos, right? People, when they sell things, they need photos of the thing. that’s where buyer decisions are made, and they pay like 35, $50 per photo professionally for these images that they’re putting on Shopify. And so people are coming up to us, they’re being like, oh I waited like six weeks for this photo set. For, 250 photos, it cost me 10 K or 5k, whatever. And we’re like oh this is interesting, right? Because, there’s this new image stuff going on and maybe we could really leverage that, to make this workflow better.So that’s really where we came up with the idea for, what is now Stylized, Stylized.Ai. What we’re building is professional product photos, for people running e-commerce shops in under 30 second

Dhaval:
oh wow. So tell us a little bit about where your product is at this stage and have you launched, is it in the pre-launch stage? Is it in the wait list, stage, et cetera, et cetera.

Boyang:
Yep. We have launched, we are soft rolling out a launch with this is a prosumer product, right? And so we’re really concentrated on the B2C motion of go to market. We’re doing like organic seo. we’re running a bit of ads on the side. And so this is all just to build up , a brand name and also to get really fast iteration on what the product surface is. So we have soft launch. We have about 200 customers right now. That’s growing probably at a rate of, I would say, like 15 to 20 per day. Which is pretty good, right? It’s only been like, I think since we opened the beta. To one and a half weeks. So we are pretty happy with that. And I think the goal for us really is to get that distribution and get in front of people into their workflows such that we get embedded. And because no one really knows, like honestly, no one really knows where the AI models are going. And so by in the next three months, someone could do some really magical stuff, right? And we want to be able to put that magic in front of customers. And to do that, you have to have a big audience. So that’s our goal, right now

Dhaval:
That’s awesome. Focusing on distribution first, that’s novel. Most of the founders and product creators, they get their heads down and they start building the product and it. They spend months and months and months before even thinking of the first interaction with the customer. And as you already know, it never goes as per the plan. Right. So what is the main value proposition? What is the main customer pain point that your product solves for?

Boyang:
Yep. Customer pain point. I have a product. I’m trying to sell it. I have a Shopify store. I need good images. Right now my options are I get like a light box set up, which is, they can be pretty complicated, right? I need to set up a photo studio area in my house. I need to take pretty meticulous pictures. I need to then learn Photoshop and edit those just the way I want them. Or I go to a professional studio and get my photos back to me, in a couple weeks. And so that’s my blocker right now. I can’t get it on Shopify. So what we do is, you. An iPhone image, and as long as it’s like pretty good neutral lighting, like anyone can do this. I’ve done this many times and I’m bad at taking photos. Right. So I do that in my product. In 30 seconds I get a virtual light box. So this is a staged, 3D rendering actually. And the technology is relatively, I wouldn’t call it simple, but it pieces together a bunch of existing models, right? To render your product in 3D. And then you get to adjust whatever you want about that rendering such that you know your product is professionally lighted. You get to change all the backdrops as if you are in a product studio or in a photo studio with like different types of materials or Hey, I want this on marble or slate, or all of that stuff. But you get to do that from the comfort of your computer and the iteration time is, on the order, five seconds versus two weeks. so that’s what we’re going for.

Dhaval:
Wonderful. Yeah. I’ve been a product owner for e-commerce companies and that finding good stylized photographs of your product has been the biggest game changer. Like experiencing, showing people experiencing the product has been the biggest game changer. So you’re solving a real pain point. You said you are in a prosumer space. If you can unpack that a little bit, why is that presumer and not just e-commerce sellers. How do you differentiate?

Boyang:
Yeah. So for us, the biggest differentiation is whether we are b2b, which would be selling to e-commerce platforms. And we’ve talked about this as well as whether we could go to Shopify and say, Hey we have an api, or we have a third party tool that. You could purchase for your sellers to make their shops more efficient. Whether that’s the route or if we want to go directly to the customers themselves. So we see that as more presumer because it’s self-serve one, we’re just launching to anyone. You can come in, you enter your email, um, you upload a picture and boom, it’s there it’s free to use. You just get these premium add-ons. And that’s how you are introduced to that product at first. So we’re calling it prosumer mostly because all of our customers are independent shop owners, and they really get to make the decision about their own product.

Dhaval:
Very cool. Yeah. So your product roadmap could be either build up your distribution, get a lot of, Prosumer e-commerce, use your product, and then become down the line, become this extension or plugin for all the e-commerce outlets that are out there. Is that something you’re thinking of?

Boyang:
Yes. I guess I won’t go into too much detail, but we do have a Shopify extension that’s coming out soon. As I said, we’re focused really on hitting that distribution and just nailing it, getting as many customers as possible. And one of the, one of the benefits here is like we’re solving one. Very individual problem, right? So it’s, Hey, I need a photo. There’s a very clear input, like I take a photo, there’s a very clear output, I get a better photo back, right? That takes 30 seconds, 15 seconds, whatever it might be. We’re solving that pain point. So it’s very easy to get in front of people and say, Hey, look. This is what you’re getting from us. and it’s easy to onboard. And then from there, I think the strategy here is if we get many customers, we can start building up, catalog extensions. We can say, Hey, put your entire store catalog with us, like we will optimize it. Or you get better photos through all of your store. And then we can really start to leverage all of the newer AI things that we see coming out in the future. So if one day there are, very good AI models for just creating catalog webpages, we could leverage that. We could then let you know our customers create their own webpages directly from our surface. And that is really the expansion route that we’re foreseeing. It’s like an asset first, website.

Dhaval:
Very cool. So we have talked about your journey. We have talked about your product. We have talked about your potential future roadmap. What I would love to dive into now is unpacking the product stack a little bit for people, for product managers or product owners who want to either make an AI product or wanna add AI to their product. What was your journey like when it comes to building this AI product, what was the stack like? Anything you can share that would really be tangible that would really help our audience.

Boyang:
Yeah. I think it’s, let’s go back to August, right? So in August my co-founder and I were Playing around with a whole bunch of ideas.One of them we settled on was this 2D textures for games. So he has a background in, in 3D games and at Facebook running Oculus. And so what we did was we just made a very simple, page with you enter something and you get back a texture, that book like that thing. so that was our first foray into any of these newer diffusion models. And from there we started. Seeing the same sort of, using the same backend, toward different customers. And this was really just week after week of iteration where we strategized and we said, Hey, what is the thing we wanna focus on? Oh we believe that workflows are the thing that really matter, right? You have to get embedded into your customer’s workflow. So if you make something where they come and they get their thing and they just like immediately leave you can get a lot of people coming to your site, but it’s not it’s not gonna grow into anything. Substantial, All right. So we targeted the workflow we started targeting the customer, which we said was e-commerce mostly for the large market. And then the actual product stack itself. We sort of device an engine to create apps very easily where it’s the top level is UI that we build with Chakra. So it’s very fast for us to just iterate on this thing.

Dhaval: What was that Chakra?

Boyang:
Chakra is a, like a UI toolkit. For React. I think it probably supports other frameworks now. And then on the backend for us, uh, we have essentially a, a lot of APIs that go to different, ML models that we’ve stood up ourselves. And so as we keep building, different services, we stand up more of these models. And then we just have access to so much flexibility at this point where we’re like, Hey, we need to upscale this image. But we already have that, right? We built this for another reason entirely, but as long as we can leverage the same flows, um, that will just work. Yeah, I think if that answered the question.

Dhaval:
Yeah, that was very helpful, Boyne. Thank you. So there has been this debate about building on top of existing ML models versus standing up your own ML models and your own ML infrastructure, which route? Should a product creator take if she wants to build an AI product, what is the best, what is the best way to test your hypothesis? And at the same time, what is the best way to actually create a high efficacy product? If you were to offer some input there?

Boyang:
I think it depends on where you think your talents and expertise are. For us, we’re both technical. It was pretty easy to set up these models on our own side. We are not training them ourselves, right?And so , we’re using open source packages, but it’s easy to set that up if you believe that your expertise is in go-to market and, SEO and you just want to set up things that provide value very quickly. But you don’t care about really optimizing the underlying stuff as long as you can get the users. Then Sure. I think that route is really good. I would just say the most important thing is for us to understand what our strategy is. And the strategy has always been, hey, like we are two technical people and we’re focusing on distribution because that’s sort of our weakness at this point. So we’re trying to short that up. And that’s why we made decisions we did. But if you are, coming from product and you just wanna stand something up these days the models are so good and there’s so much out there in terms of SaaS that just lets you, get value immediately and get APIs immediately. That I think is the better route if you were to want to iterate really fast.

Dhaval:
Very cool. So either distribution could be your advantage, or the core ML could be your advantage. You gotta figure out which path you want to take. It could be both, but you have to choose one at a time and then go forward and then iterate from there. Did I get that right?

Boyang:
Absolutely. I think for us we very early on we said, Hey, if we’re going into this ML space neither of us is a PhD, right? We’re not truly confident in our ability to produce net new models or like net new ways of doing things, however, We are very confident in generating better UX in like allowing people, and this is the whole ChatGPT argument, right? Like it’s better UX on top of something that is roughly the same and didn’t really take that much, but it was a step function, difference in value. And so that’s been our thing, like we’re gonna leverage this ux and build as fast as we can to get the most users. And then the models will come. They’ll come later.

Dhaval:
Wonderful. Now, changing gears a little bit, are. Generating revenue as of now. What is the price point? How did you come to that conclusion? Are you pre-revenue, post revenue? Tell us a little bit about your revenue stuff.

Boyang:
Yeah, pre-revenue. We’re basically concentrated on gathering users. so we have built up a, an email list, of users. And the point of this is like to get iteration as fast as possible to come to a good pricing model. And we have one in mind. We’re building it out. I think we’re gearing up for this product hunt launch that will come in like the next week or week and a half. so I think that is the point at which we will find more legitimacy in saying, Hey, we have value to offer, that you can pay for. Right now it’s a bit early and it just seems like. If we were to ask for the money, it’d be, we’re not really providing that much, so there’s not much to pay for. so that’s the revenue argument. I think also for us, if distribution is so important we don’t really wanna slow it down with anything. we would rather people see us as, Hey, we remove backgrounds really well and we give you like a flat white picture with nice shadows of your product and that’s free. And if a million people use that, then we’re very happy with where we.

Dhaval:
Wonderful. Tell us a little bit about your venture, uh, funding strategy. Are you, have you raised capital? Are you raising capital? How is it compared to your past startups in the current environment? Yeah.

Boyang:
I think funding We know the funding environment’s a bit tough. it might not be as tough as it was, like the end of last year. we are not seeking money right now. I think we want to be in as strong a position as possible while going into that fundraising. And for us it’s like we have milestones that we know we want to hit, before we really had conviction that our product is good and, you know, will have product market fit in the future if we throw money at it.. So right now, money’s not the problem. we want to get to enough users, enough retention enough growth on the payment side, in order to say, Hey, like we know now the equation is if we go seek fundraising, we get this like $3 million. Here’s what we are gonna throw it at and this is how it will help us grow.

Dhaval:
Yeah. So becoming clear on where you want to invest is where you are at right now. So do you have any funding, have you done any seed rounds or, is it all bootstrapped until this point?

Boyang:
so, last year my co-founder and I joined a, I would call it an accelerator. It’s a community, sort of like Y Combinator, it’s called South Park Commons. been really great for us as like a networking opportunity. and then that is where we got our umbrella. It’s like a pre-seed. it’s really just a parachute for us to, you know, jump off a cliff. So both of us were coming from relatively big tech jobs, so pretty cushy jobs. And, that was what gave us the confidence to just be like, Hey, let’s let’s quit. Let’s do this instead. Um, and yeah, that, that money has gotten us to where we are. We’re not really, we’re not paying ourselves very much, so we have plenty of runway. and you know, for us it’s really. Like we’re not looking for like the big investment dollars. I think it really makes us happy to just build things that people want.

Dhaval:
Wonderful. Tell us about the future vision, like really long-term future vision of, stylized.

Boyang:
Yeah, so we want stylized to be an asset first e-commerce platform. And what this means is I can imagine a future where you wanna start a shop. And you don’t go to Shopify because you need a website builder or a domain name or any of those payments options. You come to us because you have the thing in mind that you want to sell. And you can immediately launch the shop of your dreams, right? That is online. And so all of that virtual stuff, I actually don’t think, and a lot of shop owners have told us this. The boiler plate of setting up what your shop, looks like necessarily on, in virtually is part of what they want to do. They’re passionate about making good product, getting customers selling that product and like iterating on that. And that’s what makes them happy. And we want to take away like, All of the abstractions up until that point, right? So we want you to say, Hey, I have a good product, or even I have an idea for a good product. What should my pricing model be? How do you do the SEO for this? What should it look like? How do I set up a website that is visually enticing to anyone who comes in? And once we have, I think these better image models, we’ll be able to even A/B test the actual images themselves, which is something I think up till this point no one was able to do.You can’t just run to a photo studio every week and be like, Hey, give me like five sets of different photos I want to try out on different users. But what were you willing on is like, Hey, this seller likes these photos. We’ll just, on demand, we’ll generate those of your product, and present it to them. And I think that is the real e-commerce experience that we’re going for.

Dhaval:
Wow. Okay, so you’re starting with The seller’s mindset. The seller’s product mindset first, which is, I have an idea of selling this product. Help me launch a store, get a website, get the product page up, get the content up, and I can just focus on the product innovation. Did I get that right? Are you focusing on like taking over? All the abstractions, including the e-commerce, including the website, including all the marketing, including all the content. Did I get that right? Is that your vision?

Boyang:
Yeah, that’s the vision. And I think the way that it’s been working so far, in the last 20 years is we’ve been able to abstract away some of these things. As a startup founder, I didn’t have to set up my own HR ops or any I used gusto, right? And so these layers exist now and I think they don’t quite exist from the retail to e-commerce side. You can’t, you can see that transformation just about to happen and we want to be, To really transform e-commerce into something where you can say, well, like I wanna stand up an e-commerce store, and I click a couple of buttons and I tell you my idea and now I have it.

Dhaval:
Wow. I’m excited about where you’re going Boyang. And looking forward to your launch here. Good luck with your product launch. We’ll be keeping an eye on awesome things you’re working on and thank you so much for being part of this show. It’s an honor to have you on the show.

Boyang: Yeah, absolutely. It was great to be here.