DesignArena raises $7.9M for AI model taste
DesignArena's parent, Intelligence, raised a $7.9M seed led by Index Ventures to sell human A/B evaluation data to AI labs training media-generation models.
Intelligence, the company behind DesignArena, closed a $7.9 million seed round with Index Ventures in the lead and Conviction (Sarah Guo and Mike Vernal), A*, and Valkyrie joining in, TechCrunch reported August 3. DesignArena runs an “A vs. B” comparison platform where 5.3 million users worldwide rank AI-generated outputs, websites, images, and other formats, and the resulting preference data is sold to AI labs as feedback for improving media-generation models. Co-founder Grace Li said the service already generates $60 million in annual recurring revenue. The founders started the company around their 2025 college graduation while trying to solve a narrower problem: judging whether AI-generated games would actually be fun to play.
Why this is a business, not just a leaderboard
Public leaderboards tell you which model wins on average; they don’t tell a lab what “good” looks like for a specific audience or use case. DesignArena’s pitch is that large-scale, crowd-sourced pairwise comparisons are a more direct proxy for what actual users will prefer, and that labs will pay recurring revenue for a steady stream of that signal rather than building it in-house.
What it means for operators
If you’re building or fine-tuning a generative model, this is a reminder that automated benchmarks and human preference data measure different things, and benchmark scores can mislead you about real-world quality. Google’s own GA evaluation tooling for agents covers the quality/safety/grounding side; DesignArena’s bet is that pure taste and preference, especially for creative and visual output, needs its own dedicated data pipeline rather than a single unified eval harness. Worth watching whether more labs start treating that as a line item they buy rather than build.