Arcade - A New Kind of Marketplace šŸ”

Plus: CEO Mariam Naficy on building Arcade’s frontier AI-enabled commerce stack...

CV Deep Dive

Today, we’re talking with Mariam Naficy, Co-founder and CEO of Arcade.

Arcade is a new kind of marketplace—an AI-to-physical product platform designed to bring personalization and creativity back into online shopping. Founded by serial entrepreneur Mariam Naficy (Eve.com, Minted) and incubated out of Heretic Ventures, Arcade allows anyone to design, customize, and purchase real physical products generated from AI imagery. From accessories to home decor, each item is manufactured on demand through a global network of vetted makers, powered by fine-tuned image models and real-time pricing and detection models.

At its core, Arcade is building a bridge between the virtual and physical worlds. Using frontier models like OpenAI’s gpt-image-1, users can now edit product designs directly from the site’s interface, turning those creations into purchasable SKUs with a single click. The platform’s proprietary tech stack dynamically prices items based on what’s detected in the image—removing the need for quotes, wait times, or manual workflows. For creators, Arcade is also tool-agnostic: images from Midjourney, KREA, and other platforms can be imported and turned into real, manufacturable, monetizable products instantly.

In March 2025, Arcade announced a $25m Series A round, bringing its total fundraise to $42 million. Current investors now include Reid Hoffman, Offline Ventures (Brit Morin), Sound Ventures (Ashton Kutcher), Canaan Partners (Laura Chau), Forerunner Ventures (Kirsten Green), Inspired Capital (Alexa von Tobel), and Torch Capital (Jonathan Keidan). New angel investors Sara Beykpour, Kayvon Beykpour, Anna Veronika Dorogush, Eugenia Kuyda and Marissa Mayer joined existing angel investors Christy Turlington Burns, Colin Kaepernick, Karlie Kloss, David Luan, and Jeff Wilke.

In this conversation, Mariam shares how she built Arcade’s frontier commerce stack, why pricing is the unlock for AI-generated products, and why the most important infrastructure problem in AI right now might actually be supply chain.

Let’s dive in āš”ļø

Read time: 8 mins

Our Chat with Mariam šŸ’¬

Mariam, welcome to Cerebral Valley! First off, introduce yourself and give us a bit of background on you and Arcade. What led you to co-found Arcade?

Hey there! My name is Mariam and I’m the Cofounder and CEO of Arcade. 

A bit about me: I’ve been an entrepreneur for many years—I've started 3 companies. The first was back in 1998: the world’s first online cosmetics retailer. It was very early—before clothes or shoes were even being sold online. It was called Eve.com, and I sold it in 2000, right before the dot-com bubble burst, to LVMH (Louis Vuitton MoĆ«t Hennessy).

Since then, I’ve focused on e-commerce and marketplaces, always building new consumer experiences. In 2007, I started Minted, one of the early leaders in crowdsourcing. We built it into a several hundred million dollar business over 15 years, supporting a community of around 20,000 artists globally. These artists use Minted as a platform for self-expression and income—submitting designs that the company transforms into products. One of the things I learned at Minted (and that serves me now at Arcade) is how to build a global on-demand manufacturing process, network and prices.   Minted is especially known for holiday cards, wedding invitations, and premium printed artwork. It’s a high-quality, premium brand with a really strong market share, and I’m incredibly grateful to our customers for believing in us and for their loyalty. I now serve as Chairman of the Board.

I left to start a venture incubator called Heretic Ventures, which was funded by Reid Hoffman, Ev Williams, and a few others. The goal was to launch more consumer internet businesses. But then the idea for Arcade came to us, and as a team, we felt we had to stop what we were doing at the studio and focus entirely on it. It was a great opportunity, and we believed that being hands-on—actually building something ourselves—would make us better investors and incubators in the long run. 

So, we all shifted over and started Arcade. We officially launched last September, but had been working on building our supply chain and proprietary volumetric and manufacturing data for about a year and a half before that.

How would you describe Arcade to a creator or customer who hasn’t heard of it?

Arcade’s mission is to create super-memorable shopping experiences and bring a sense of inspiration back into e-commerce. That’s where the name ā€œArcadeā€ comes from. In 1800s Paris, shopping arcades were places people would escape to, away from the chaos of the streets. But we also liked the gaming implication of ā€œarcadeā€ā€”a place where you can play and collaborate around products.

That spirit is core to what we’re building. Arcade is the first idea-to-reality physical product marketplace—the first AI-to-physical product destination. Our goal is to launch across many categories. We started with accessories and recently expanded into home decor. Accessories is a $350 billion market in the U.S., and home decor is around $750 billion. Both are highly expressive categories, but the online shopping experience for them has become overwhelming. At some point, the fun of the hunt wears off—whether you're looking for the perfect home piece or a piece of design that really reflects who you are. We want to solve that.

The intersection of AI and commerce is a vast space that feels ripe for innovation. What inspired you to focus on this area at the start of your journey with Arcade?

When I first saw DALL-E—this must’ve been late 2022—I immediately thought, people are going to want physical artifacts of the things they’re creating. So I figured I might as well skate to where the puck is going and be the first to do it. I saw a huge opportunity for AI to permeate the physical world, and I was excited by the idea of being the bridge between the virtual and the physical.

I already had a lot of experience with making custom products on demand. We built a supply chain at Minted to deliver personalized products on time for Christmas, and at one point, we had one of the biggest API-connected print networks in the country. It could scale massively, especially during the holidays. So I was thinking, how do I do that again—but this time with AI?

The answer was to build a network of independent makers and manufacturers around the world who could produce these high quality products, and actually enhance their businesses by doing so. Now, consumers can communicate exactly what they want using images or text. They can say, ā€œThis is the piece of furniture I want—can you make it?ā€ We train models using data provided by these makers to ensure manufacturability.

But what really compelled me was the idea that anyone could become a creator. Anyone who’s been shut out of the product design process—whether it’s because of cost, access, or resources—can now participate. Solopreneurs, hobbyists, artists... anyone. I saw this as a massive vehicle for human empowerment, and I’ve always been drawn to businesses that unlock that kind of creativity and support.

What are some of the key markets you’re excited to venture into, now that Arcade’s technology can be applied to a wide variety of products? 

I think just getting through home decor alone is going to take us a while this year, but we’re rolling things out pretty quickly. For example, we’ll be adding scarves to the accessories category soon—really artistic, high-quality pieces.

On the home decor side, categories like glassware, textiles, and other products will be launching rapidly. A lot of our textile products are rolling out over the next month, and the plan is to continue building through all of the home decor category this year.  We’d like to listen, too, to our community, and see where they’d like to take us.

Talk to us about the technical innovations behind Arcade. What are some of the model architectures and innovations you’re leveraging under the hood to power Arcade? 

The API for GPT’s image-gen came out on May 24, and we integrated it into our live site just a few days later. As far as I can tell, this is one of the first times a frontier model like this one has been deployed live in a consumer commerce setting—not just in prosumer or creative workflows. 

If you visit our homepage, you’ll notice we’re running an A/B test. The UX you see now isn’t final; it’s an intentional experiment. 

When you hover over a product card, there’s now an ā€œEdit with AIā€ button powered by GPT image-gen. It’s fully live. And so far, the interface is working really well. You can take a physical product and tailor it however you like.

The key innovation on our side is pricing. We’ve developed proprietary data pipelines to fine-tune image models, but more critically, we’ve built detection models that analyze what’s in an image and assign a price in real time. That’s the breakthrough. There’s no magic if there’s no price—if you still have to wait for a manual quote, it’s just a design tool, not a commerce engine.

What makes this magical is that the edited product immediately becomes a SKU—something you can buy, just like pulling it off a shelf. Behind the scenes, this required getting our supply chain partners aligned on dynamic pricing algorithms and making sure those work hand-in-hand with our detection models. So if you take an item on our site, edit it with gpt-image-1, OpenAI returns the modified image, and we price it instantly.

Talk to us about how you’ve structured your supply chains to deal with the spikes in demand from digital-to-physical product demand? That’s a serious engineering problem to tackle in the midst of a technology shift.   

On Arcade, once you check out, everything moves instantly. You prompt your image, edit it, and select your ring size or other details—just like a standard product. At that point, it becomes a real SKU in our physical product database. From there, it’s like any normal e-commerce experience: pick your size, add to cart, and check out. The order immediately goes to the manufacturer and production begins.

Before it ships, you’ll receive a video of the finished product. If you don’t like it, you can give feedback or message the maker directly, though that part’s optional and not built into the default flow.

The hard part is everything under the hood. Building the supply chain, finding the right high quality makers, designing pricing algorithms for each product line, and getting buy-in from all stakeholders. These pricing algorithms are deeply connected to our detection models, so whatever we can detect in the image directly influences the final price.

It’s this blend of technical infrastructure and business design—understanding how manufacturing works, what drives cost, and translating that into pricing logic that’s both accurate and scalable. That’s the heart of what we’ve built.

Pricing is a huge piece of e-commerce - how are you approaching the complex question of pricing for AI-generated digital products?  

It feels totally sci-fi to users—like something out of The Jetsons—that you can take an image, modify it, and instantly get a price on the physical product. It’s so new that people react in really different ways. Some consumers just assume AI can do anything, like it’s literal magic. They see it and go, ā€œOf course that’s possible,ā€ and have very high expectations. On the opposite side of the spectrum, other consumers don’t really can’t believe it’s possible because it’s so new and different.

But building this is hard. You have to train the pricing and detection models, stitch them together, and align the entire supply chain around how it all works. That’s where the real magic is—instantly pricing a physical product from an AI-generated image.

Tell us about the team - how would you describe the culture at Arcade? Are you hiring, and what do you look for in prospective team members joining the team? 

We’re always hiring, especially in AI engineering and AI product. If that’s your background, we’d love to talk. This is probably the best startup team I’ve ever worked with—we’ve built something really special here. The bar is very high, and what we look for most is deep intellectual horsepower, a serious commitment to the craft, and a learning mindset.

Our Head of AI has a PhD in math and academics and specializes in end-to-end ML pipelines for multimodal systems, production-grade AI deployment, and innovating across text, image, and language models. Our head of supply chain has 20+ years of experience across home decor, fashion, and operations at places like Coyuchi and Gap. Across the board, we’ve brought together humble, low-ego people with exceptional domain expertise. We’re about 18 people in total right now. 

Anything else you'd like our readers to know about your vision for Arcade?

Technically, one of the more interesting things is that we’re able to price any image—whether it’s a piece of home decor, a design, or something else—even if it wasn’t generated through our system. So creators using Midjourney, KREA, or any other tool can import images directly into Arcade, and we can turn them into live, purchasable SKUs with assigned manufacturers. You don’t have to use our generative models. Just bring your image, we’ll run it through our manufacturability models to make small adjustments if necessary, and instantly price and list the product.  That image-to-image functionality will be launching in one week and will be discoverable as a button on your gallery page after you sign up.

That’s one of the things I’d really emphasize: if you’re a creative using generative tools today, Arcade gives you a direct path to turn your ideas into real, physical products that anyone can buy!

Conclusion

Stay up to date on the latest with Arcade, learn more here.

Read our past few Deep Dives below:

If you would like us to ā€˜Deep Dive’ a founder, team or product launch, please reply to this email ([email protected]) or DM us on Twitter or LinkedIn.