Intelligence raises $7.9M seed to scale human taste data for AI

Original: Design Arena creators raise $7.9 million to bring taste to AI models

Why This Matters

Human preference data is becoming a paid commodity for frontier AI labs seeking to improve design-quality outputs at scale.

Intelligence, the company behind Design Arena, announced a $7.9M seed round led by Index Ventures on August 3, 2026. The platform has 5.3 million users and generates $60M in ARR by providing human preference rankings to frontier AI labs for design and media evaluation.

Intelligence, founded in 2025 by Grace Li and college co-founders, announced a $7.9 million seed round led by Index Ventures, with participation from Conviction (Sarah Guo and Mike Vernal), A*, Valkyrie, and others. The company originally set out to build an AI game engine but pivoted after realizing that scalable human feedback was a critical bottleneck for improving AI-generated design outputs.

The resulting product, Design Arena, now serves 5.3 million users globally. Consumer users interact with a ChatGPT-style interface that lets them compare AI-generated outputs across formats including websites and images via an 'A vs. B' ranking system. On the enterprise side, frontier AI labs pay to access this continuous stream of human preference data to improve their media-generating models.

Li states the platform is currently generating $60 million in ARR. Because users must log in to receive outputs, Intelligence can also track how aesthetic preferences vary across regions and over time — for example, noting that web dashboards in Asia tend toward more maximalist design styles.

The company positions its human evaluation data as a complement to automated benchmarks, which are susceptible to manipulation, as highlighted by a recent Hugging Face breach. However, the space carries risk: Yupp, a similar startup backed by $33M from a16z's Chris Dixon, shut down earlier this year despite having over 1.3 million users and frontier model customers.

Source

techcrunch.com — Read original →