Design Arena, a 10-person team running a head-to-head AI-output voting product, scaled

A 10-person team just turned free AI voting into a $60M ARR business in six months. Design Arena runs a simple product: users compare two AI-generated outputs, pick the better one, and move on. Five and a half million people did exactly that before the company raised a $7.9M seed led by Index Ventures.

The consumer side is free. The real product is what those votes generate: human preference data. Frontier AI labs buy that data to train their models on what real people actually prefer, not what engineers think they should prefer. Design Arena scaled from $5M to $60M ARR in half a year by selling that signal back to the companies building the models.

This is a two-sided model that works because the free product creates the valuable asset. Users get a quick, entertaining way to interact with AI outputs. Labs get ground truth on human taste at scale. The company sits in the middle, capturing value from both sides without heavy infrastructure or a large team.

If you’re running a business that generates data as a byproduct, this is worth studying. The pattern here is turning user activity into a dataset someone else will pay for. It’s the kind of insight we build into an AI command centre so you can spot where your own operations might be creating similar assets without realizing it. Ten people built this in months because they understood what the data was worth before they built the interface.