Grace Li Design Arena is a founder story that began as an unglamorous side problem inside a college project that never worked out. Li is the co-founder and CEO of Intelligence, the company behind Design Arena, a crowdsourced benchmark that measures whether AI-generated design, websites, images, video, and full applications, is actually good, not just functional. On August 3, 2026, the company announced a $7.9 million seed round led by Index Ventures, revealing along with it a growth number that stopped people mid-scroll: in six months, with a team of just 10 people, Design Arena scaled from $5 million to $60 million in annualized revenue.
Here is how a college side project about AI-generated video games became one of the AI industry’s fastest-growing evaluation platforms.
Who Is Grace Li?
Grace Li is the co-founder and CEO of Design Arena, according to her own personal website and the company’s Y Combinator profile. She studied computer science and neuroscience at Harvard University, graduating in 2025, and was a student-athlete on Harvard’s varsity sailing team during her time there.
Before founding Design Arena, Li worked at Apple, contributing to Photos, Memory Creation, and Apple Intelligence, work significant enough that she was selected to demo a project live to Craig Federighi, Apple’s senior vice president of software engineering. That kind of direct, high-visibility product experience inside one of the world’s most design-obsessed technology companies shaped the instinct that would later become Design Arena’s entire premise: that good design is measurable, even when it’s hard to define in words.
From Apple’s Product Teams to a College Game Engine Project
Li co-founded Design Arena with Kamryn Ohly, who studied computer science and education at Harvard and also previously worked at Apple, describing themselves and their team as “all best friends from Harvard.” The company’s origin traces back to a few weeks before Li’s 2025 graduation, when she and a handful of college friends were trying to get an AI-powered game engine to work.
The models could generate functional games, but none of them were actually fun to play, a gap that raised a harder question than the team initially expected: how do you even measure whether a game is fun in the first place? Li and her co-founders concluded there was no substitute for real human judgment, and began looking for ways to gather honest human feedback on AI-generated output at scale, an idea that would become Design Arena within weeks.
The Problem Design Arena Set Out to Solve
Large language models can pass bar exams and win gold medals at international math competitions, but Li and her co-founders noticed they still routinely generated interfaces with basic aesthetic failures, like white text on a white background, because no standardized way existed to measure taste, usability, or design quality. Without a benchmark, Li has argued, there was no real pressure on AI labs to improve that specific capability.
That gap mattered more as AI-generated design output, websites, marketing images, app interfaces, began shipping directly to real users, where a technically functional but poorly designed product still fails in the market. Design Arena set out to become the missing scoring layer AI labs needed to actually track and improve design quality over time.
Building Design Arena: Crowdsourced AI Evaluation at Scale
Design Arena works like a sophisticated model router paired with a head-to-head ranking system. Users submit a design request, a website, image, video, or one of a dozen other formats, and the platform generates competing outputs from different AI models, then presents users with a series of “A versus B” comparisons until the outputs are ranked from best to worst using an Elo-based scoring system similar to competitive chess rankings.
That structure serves two audiences simultaneously: individual users get a genuinely useful tool for generating and comparing AI design output, while participating AI labs get a continuous stream of real, organic human feedback they can use to improve their own models’ design capabilities. Li has described the platform’s real commercial value coming specifically from that enterprise side, where frontier AI labs treat Design Arena as an ongoing source of feedback data rather than users treating it as a one-time creative tool.
Inside Design Arena’s $7.9 Million Seed Round
Grace Li Design Arena’s parent company, Intelligence, announced a $7.9 million seed round on August 3, 2026, led by Index Ventures, with participation from Conviction, the venture firm founded by Sarah Guo and Mike Vernal, alongside A*, Valkyrie, and continued backing from Y Combinator. The company says it will use the funding to build core machine learning evaluation infrastructure, expand its crowdsourced reviewer community, and scale its data engineering pipelines for multi-modal design validation.
That relatively modest seed size, compared to the company’s already-substantial revenue, suggests Design Arena raised primarily to accelerate infrastructure and community growth rather than to fund a pre-revenue product still searching for its first paying customers.
Growth and Traction: $5M to $60M ARR in Six Months
Design Arena says it scaled from $5 million to $60 million in annualized recurring revenue in just six months, with a team of only 10 people, reaching more than 5 million users across more than 190 countries. Li has described the company’s path to its first major enterprise deal as remarkably fast: roughly a week after realizing frontier AI labs would pay for the kind of feedback data Design Arena generates, the company closed its first deal with a frontier lab.
“It was the missing bottleneck for a lot of these models to make improvements in the design space,” Li said, describing why AI labs proved so willing to pay for access to organic, at-scale human feedback rather than trying to build equivalent evaluation infrastructure internally.
The Investors Backing Grace Li
Index Ventures leading Design Arena’s seed round, alongside Conviction’s participation from AI-focused investors Sarah Guo and Mike Vernal, signals investor conviction specifically in the company’s position as evaluation infrastructure for the broader AI industry, rather than treating it as a narrower consumer design tool. Continued backing from Y Combinator, which the company graduated from before this round, reflects sustained investor confidence from a source with direct visibility into the company’s progress since its earliest days.
That combination of AI-native investors rather than generalist consumer-technology firms suggests the round was priced and evaluated specifically around Design Arena’s role feeding data back into frontier model development, a distinct investment thesis from a typical consumer AI application.
Design Arena’s Position in the AI Evaluation Race
Design Arena occupies a specific, increasingly important niche within the broader AI industry: benchmarking and evaluation infrastructure, rather than building foundation models or consumer-facing AI applications directly. As AI coding agents and design tools mature, the company has pointed to evaluation itself becoming a core bottleneck, arguing that testing models against real-world application-building tasks, rather than narrow, artificial benchmarks, is what actually reveals where model capability still falls short.
That positioning gives Design Arena a business model that scales alongside the broader AI industry’s growth rather than competing directly against any single foundation model provider, since frontier labs across the industry represent potential customers for its evaluation data rather than competitors to its core product.
What’s Next for Design Arena
With fresh seed capital, Design Arena’s near-term priorities center on building out core machine learning evaluation infrastructure, growing its crowdsourced reviewer community, and expanding its data engineering pipelines to support multi-modal design validation beyond its current website, image, video, and full-stack application benchmarks. The company’s continued publication of live leaderboards and research notes suggests an intention to maintain public credibility as a genuinely independent benchmark, rather than one perceived as favoring any single AI lab.
Li’s own team has framed Design Arena’s mission as part of a broader effort at parent company Arcada Labs to understand and distill human experience more generally, suggesting ambitions that may extend beyond design evaluation into other domains where human judgment remains difficult for AI systems to replicate on their own.
Lessons for Entrepreneurs From Grace Li’s Journey
Li’s path offers a few clear lessons for young founders building a company directly out of a failed side project. First, Design Arena emerged from honestly confronting why an earlier project, the AI game engine, wasn’t working, rather than abandoning the effort entirely once the original idea stalled. Second, recognizing that the company’s most valuable customers would be enterprise AI labs rather than individual consumers, discovered only after users started organically generating valuable feedback data, shows how a product’s real business model sometimes reveals itself only after launch rather than being fully planned in advance.
Third, Design Arena’s extraordinarily lean team of 10 people supporting $60 million in annualized revenue suggests a deliberate focus on capital and team efficiency from the earliest days, a discipline that gave the company leverage in fundraising conversations most 10-person startups would not have.
Frequently Asked Questions
Who is Grace Li? Grace Li is the co-founder and CEO of Design Arena, an AI design benchmarking platform. She studied computer science and neuroscience at Harvard University, graduating in 2025, and previously worked at Apple on Photos, Memory Creation, and Apple Intelligence.
What does Design Arena do? Design Arena is a crowdsourced benchmark that evaluates AI-generated design output, including websites, images, video, and full-stack applications, using head-to-head comparisons and an Elo-based ranking system to measure design quality and usability.
How much funding has Design Arena raised? Design Arena’s parent company, Intelligence, raised a $7.9 million seed round announced August 3, 2026, led by Index Ventures, with participation from Conviction, A*, Valkyrie, and Y Combinator.
How fast has Design Arena grown? Design Arena says it scaled from $5 million to $60 million in annualized recurring revenue in six months, with a team of only 10 people, reaching more than 5 million users across more than 190 countries.
Who co-founded Design Arena with Grace Li? Design Arena was co-founded by Grace Li and Kamryn Ohly, who serves as chief technology officer. Both studied at Harvard University and previously worked at Apple.
Where is Design Arena based? Design Arena’s parent company, Intelligence, is based in San Francisco, California.
Conclusion
Grace Li Design Arena is a reminder that some of AI’s most commercially significant infrastructure companies emerge from founders honestly confronting why their original idea didn’t work, rather than from a carefully planned business strategy. With $7.9 million in fresh seed funding, revenue growth from $5 million to $60 million in just six months, and a lean 10-person team serving both individual users and frontier AI labs, Design Arena is testing whether teaching AI models good taste can become as foundational to the industry’s next phase as raw capability benchmarks have been to the current one.
For more founder stories from young, fast-moving founders and women building in AI, see Jenny Duan’s Clair Health, Julie Bornstein’s Daydream, and Nathan Arnold’s Photon Queue on Denote Press.
