TechCrunch: Intelligence raises $7.9M for DesignArena human-taste platform

Intelligence, the company behind the design evaluation platform DesignArena, has raised a $7.9 million seed round led by Index Ventures, with participation from Conviction (Sarah Guo and Mike Vernal), A*, and Valkyrie. The announcement was made on August 3, 2026.

DesignArena lets users compare AI-generated visual outputs side by side, rating websites, images, and other design assets in simple A-vs-B format. That stream of choices becomes training signal: frontier AI labs use the human feedback to improve model performance specifically in the visual domain. Co-founder Grace Li and her colleagues started the venture in 2025 after noticing that their AI game engine could produce functional games but none that were actually fun — which led them to build scalable, honest human feedback as a dedicated product.

The platform has reached 5.3 million users worldwide and $60 million in annual recurring revenue. It also tracks preference variation across geographies and over time, giving labs both aggregate taste data and insight into how design preferences shift by region and era.

Why it matters for designers. The current gap in AI visual quality is partly a data problem: models trained on objective benchmarks can be gamed, while automated metrics have no sense of aesthetics. DesignArena bets that scalable human evaluation can fill that gap, and its revenue numbers suggest real demand without requiring marketing spend.

For designers, this signals a broader shift: human judgment about visual quality is starting to be treated as structured, tradeable data rather than an unquantifiable soft skill. The work of deciding which design is better — something designers do routinely — is now infrastructure for the next generation of AI tools.

The competitive landscape is uneven. Yupp, a comparable human-feedback startup, failed despite $33 million in funding; LM Arena, focused on language model evaluation, raised a $150 million Series A. DesignArena sits in between, specific to the design domain, with revenue that suggests it found product-market fit early.

The seed capital will go toward expanding the platform and deepening enterprise relationships with AI labs that need design-quality training data at scale.