The problem that Gradient is trying to solve seems a little strange. It’s not meant to be rude; rather, it’s more like how the best startup ideas often feel obvious after someone else has already found them. Tim Clothier had around 7 million collectible cards. They took up most of his garage and into his living room. From some seats, they even blocked your view of Mount Rainier. He knew how long it would take to sort them by hand: about 25,000 cards a week, over fifteen years of weekends. His wife knew. His friends knew. Most of them changed the subject in a quiet way. Matt Lubbers asked a different question.
It’s been more than four years since that first talk. They are now co-founders of Gradient, a company based in Renton, Washington, that uses custom robots and AI to sort, catalog, photograph, price, and list trading cards for collectors like Clothier who have more cards than time. The office of the company is right across the hall from the offices of Seattle Sounders FC. This seems like a good fit for a company that deals with both serious technology and die-hard fans.
The cards are the first thing you notice when you walk through the Gradient workspace. Every day, boxes of them from all over the country come in the mail and are stacked on rolling racks that run along the walls. The building has space for three times as many cards as there are present (10 million). You can open any box and find baseball heroes, Pokémon characters, and football legends. It’s all the stuff that Americans have been collecting since the 1980s but have forgotten about. People sort cards by hand around a few tables. Some people write code at computer stations. There are parts of this place that make it feel like a nerd’s dream—parts card shop, part engineering lab.
Eight robotic sorters hum and pulse in one corner. There were short bursts of air, mechanical whirring, and cards moving through a custom rigging system that Lubbers, the company’s CTO, made himself. He worked for fifteen years at companies like Amazon Robotics and Zipline, where he built systems for self-driving cars and drones that could fly themselves. Here, robots use soft suction fingers to pick up cards and move them across flatbed scanners to take pictures of the backs. At the same time, cameras on the ceiling take pictures of the fronts. There is a physical and digital catalog of every card. Every day, the machines can handle up to one hundred thousand cards. Lubbers wasn’t ready for photographers to get close to them yet, which shows how much he cares about what he’s made.
The images are sent to a nearby server room that has three custom supercomputers. Each of these computers has six GPUs set up in a way that is similar to how NVIDIA’s high-end chips are set up. Every day, the system looks at 500,000 pictures and compares each card to a database of 30 million different ones. You can’t call it a small business, even though the company is still young.
This group has raised $6 million, mostly from family and friends. Adrian Hanauer, who owns most of the Seattle Sounders and has known Clothier since he was 15, is one of the investors. Twenty-five people work for the company, which is just starting to get customers from outside the area. The largest one so far has sent over 500,000 cards to be processed. Prices for subscriptions range from $9.99 a month for casual collectors to $99.99 a month for business users. It costs 40 cents per card to scan as you go. Based on the level, Gradient takes between 16 and 20 percent of the sale price of a card.
It’s important to note that Gradient isn’t the only one in this area. Collectors can already use their phones to price cards on sites like CollX, Ludex, and eBay’s own “scan-to-list” feature. The most likely industry comparison is TCG Machines, which is based in Calgary and has robotic sorters that work in 20 countries and have cataloged close to a billion cards. Gradient is going after a market worth about $15 billion in the U.S. alone. This market has been getting tech attention for a few years now. There is still a lot of uncertainty about whether Gradient can find a stable place in that landscape. The robots are really cool. There are still not many customers.
Still, there’s something appealing about the first idea here. Someone is probably staring at a shoebox full of cards they haven’t touched in twenty years right now, and they have a vague idea that some of them might be worth something. Gradient is putting its money on that person and on the idea that computers can do in seconds what people could not do by hand for fifteen years.
