As co-founder Grace Li tells it, the company took shape just weeks before graduation in 2025, when a small circle of college friends set out to make their AI game engine work. The models could generate playable games, but none of them were actually fun—which sparked a tricky question: how do you know if a game will be enjoyable before people play it? They concluded there was no real substitute for human judgment, and quickly started brainstorming ways to collect honest feedback from real people at a large scale. That idea evolved into DesignArena, an AI tool now used by 5.3 million people worldwide. As it happened, plenty of AI companies were hunting for scalable user feedback—and many were happy to pay for it. “It was the missing bottleneck for a lot of these models to make improvements in the design space,” Li says. “About a week later, we closed our first major deal with a frontier lab, and the rest is kind of history.”

On Monday, the company behind DesignArena—called Intelligence—announced a $7.9 million seed round led by Index Ventures, with participation from Conviction (Sarah Guo and Mike Vernal), A*, Valkyrie, and others. For everyday users, DesignArena feels a lot like a sophisticated model router. There’s a ChatGPT-style prompt window, plus dropdowns for websites, images, and a dozen other visual formats. After you enter your request, format, and style, you’re walked through a series of “A vs. B” comparisons until you’ve ranked the handful of outputs from best to worst. It’s a handy tool on its own, but the platform’s real value lies on the enterprise side, where participating models can tap into an endless stream of instant feedback for their media-generating systems. Users generally don’t care which models they’re ranking—as Li puts it, they just want the best output possible—so their choices offer critical insight into what people actually want. For frontier labs, that’s a service worth paying for, Li says, noting the site is currently pulling in $60 million in ARR and cementing its role as a major source of human-led evaluation data for the AI industry.

Crucially, users must log in to get their results, which means Intelligence can also track how preferences shift across continents and over time. (Li notes that web dashboards in Asia tend to lean toward a more maximalist design style.) These human-driven measures are an important complement to automated benchmarks, which can run at much larger scale but are often vulnerable to gaming or manipulation—as the Hugging Face breach demonstrated so vividly last week. That said, crowdsourced human feedback isn’t guaranteed to be a winning market. Less than a year after launching, Yupp shut down earlier this year after raising $33 million from a16z crypto’s Chris Dixon. It had also landed some frontier models as customers and claimed over 1.3 million users, yet still couldn’t build a sustainable long-term business. Even so, other startups focused on human evaluation appear to be thriving. LM Arena, which takes a similar approach to text-based responses, raised $150 million in a Series A in January, just four months after formally launching its paid product.