EV Atlas — every Emergent Ventures grantee, in one place

Greg Kamradt's grant in cohort 41 (March 2025) was simply tagged 'measuring AI progress' — but the project behind it is one of the more important benchmarks in the AI field. He runs the Arc Prize, the public competition built around François Chollet's ARC-AGI evaluation, which measures whether AI systems can solve simple visual puzzles that humans find trivial but frontier LLMs historically struggle with. The ARC-AGI benchmark sat largely unsolved for years until late-2024 reasoning models started making real progress on it, and the Arc Prize Foundation has been at the center of tracking and explaining that progress — including running multi-million-dollar competitions to push the state of the art. The EV grant landed at a moment when 'measuring AI progress' was suddenly the most contested question in the field. Frontier model evaluation went from niche to existential almost overnight, and Kamradt's project is one of the public-facing pieces of infrastructure that the field actually relies on.