Altered deploys robots into the messy reality of rag houses to collect data and train models, before pushing that capability across the value chain. Soft-body manipulation has challenged roboticist’s for decades. Watch a robot try to fold laundry and you'll see handling clothes is nowhere near a solved problem. Altered's task is to pick, inspect and grade garments — a manipulation-and-classification problem specific to apparel. You can't simulate fabrics, so the only path is collecting real-world data. Altered gets paid to collect this data by solving a real labour bottleneck in rag houses, where failure is cheap, supply is near-infinite, and diversity is extreme.

The manipulation head start is just the entry point. The durable win is the grading data flywheel and the move up the value chain. In the US alone, 17 million tonnes of textiles are discarded every year, making rag houses a ~$2bn US market that pays Altered to collect a near-perfect dataset in a near-perfect environment. The same autonomy layer scales to resale grading ($5bn), returns processing ($30bn), and eventually custom apparel and advanced manufacturing.

Allen Cheng conceived of Altered in his mid-twenties, but he's been working on both ends of it since he was a teenager — shipping autonomy into messy real-world environments, and flipping limited-edition sneakers and vintage wares on eBay. He studied robotics at Carnegie Mellon and interned at NASA's Jet Propulsion Lab before four years on the founding autonomy teams at Anduril and Applied Intuition. Allen walked away from roles and equity that other engineers would kill for to start Altered.

At scale, Altered's technology has the potential to automate the entire value chain, changing the underlying economic structure of the textile industry. Apparel will be on-demand: what you want, when you want, where you are, built to last.