Design Arena, a platform that crowdsources human feedback for AI model evaluation, has raised $7.9 million in Series A funding. The round was led by Lightspeed Venture Partners, with participation from existing backers.

The platform serves as a marketplace where AI developers submit tasks for human evaluation. Users worldwide complete these tasks and provide ratings that help frontier AI labs understand how their models perform on real-world problems. Design Arena currently operates with 5.3 million active evaluators globally.

The company addresses a fundamental gap in AI development. Training and evaluating frontier models requires enormous amounts of human feedback. Outsourcing this work through distributed platforms lets labs access diverse perspectives and reduces costs compared to building internal evaluation teams.

Design Arena competes in a growing market. Anthropic, OpenAI, and other frontier labs have built some evaluation capabilities in-house, but many still rely on external platforms for scale. Companies like Scale AI and Surgo also provide human feedback services, though Design Arena focuses specifically on design and usability evaluations rather than safety labeling.

The fresh capital lets Design Arena expand its evaluator base and build tools for AI labs to better structure feedback collection. The platform plans to deepen partnerships with frontier labs and improve its matching algorithm to assign evaluations to users most qualified to judge them.

Human feedback remains essential for modern AI development. While automated metrics measure things like accuracy or latency, human evaluation captures subjective qualities like usability, clarity, and cultural relevance. As models become more capable, the need for sophisticated human feedback systems grows.

Design Arena's model distributes evaluation work globally, which offers advantages beyond cost. Diverse evaluators from different regions and backgrounds catch issues that homogeneous teams miss. This geographic distribution also creates employment opportunities in markets where remote work pays substantially better than local alternatives.

The startup's growth reflects confidence that human-in-the-loop evaluation