Grace Li co‑founded the company weeks before graduating in 2025, alongside college friends trying to make an AI game engine work. The models could build functional games, but none were fun – raising the question: how do you measure fun?

Realizing there was no substitute for human judgment, the team set out to capture honest feedback at scale. The result was DesignArena, now accessed by 5.3 million users worldwide who rank outputs in “A vs. B” pairings.

For enterprise clients, the service becomes an endless source of instant feedback for media‑generating models. Users care only about the best output, not which model produced it, allowing rankings to reflect true user preferences.

Why This Matters

BozokMedia analysis shows that human‑led evaluation fills gaps left by automated benchmarks, especially after high‑profile breaches like the recent Hugging Face incident. Scalable feedback not only accelerates model improvement but also tracks taste shifts across continents.

“Without real‑world user feedback, AI model development stalls at simulation,” says AI researcher Dr. Anita Sharma.

The platform’s $60 M ARR validates that frontier labs are willing to pay for reliable human data, even as competitors like Yupp faltered despite $33 M raised. Meanwhile, peers such as LM Arena have secured $150 M, proving the market’s appetite for human‑centric evaluation.

Frequently Asked Questions

Who are DesignArena’s primary customers? AI research labs, game‑design studios, and large enterprises developing media‑generation models.

Is there a free tier? Individual users can access limited features for free; enterprise plans unlock full feedback loops and analytics.