/
#94
General Intuition
World-model and spatial-temporal reasoning lab, competing in the agent-foundation-model space. Spun out of gaming-clip platform Medal, it trains agents on 2 billion first-person gameplay videos per year, using DIAMOND, a diffusion-based world model that predicts future frames directly rather than compressing them into tokens. Near-term targets: adaptive NPCs and search-and-rescue drones.
Links