Factory

Factory started as a way to share music with friends. Behind it was a bigger question: could a useful, lightweight product help people build a portable record of their taste?

2024-2025

Most recommendation systems learn about you inside closed platforms. We wanted to explore something different: a record of taste that belonged to the person and could move between products.

Starting with infrastructure would have made that idea difficult to explain—and expensive to validate. Music gave us a simpler way in. People already share recommendations, write about what they love, and build an identity around their taste.

The first version was deliberately simple: choose a track, add a short review, and share it. This let us test the behaviour before committing to the larger technical idea. I designed and built the early web product, then used what we learned to shape the native app and the infrastructure underneath it.

Factory needed to feel quick enough for a passing recommendation, but substantial enough to become a useful record over time.

I worked across the interaction model, visual design, product architecture and implementation. That included search and metadata, writing and publishing reviews, personal collections, social discovery, and the differences between web, desktop and mobile.

Around 3,000 people used Factory to catalogue more than 200,000 tracks. More importantly, it gave us a practical way to test the larger idea. People would contribute useful preference data when doing so created immediate value for them—not because they were asked to maintain an abstract profile.

That became our approach to new ideas: find the smallest useful product that could test the biggest assumption.