General Intuition

General Intuition is a New York based AI research lab building large action foundation models that aim to give agents spatial and temporal reasoning, the ability to understand how objects and entities move through space over time. The lab spun out of Medal, a platform for uploading and sharing video game clips, and is using that trove of gaming videos to train foundation models and AI agents that can reason about how things move through space and time. The target applications stretch well beyond games, with the team pointing to in-game bots and non-player characters in the near term and search-and-rescue drones and other embodied systems further out. The data thesis is the differentiator. Medal's dataset spans roughly 2 billion videos per year from 10 million monthly active users across tens of thousands of games , captured from first person play across an enormous variety of virtual environments. Because gamers who upload clips tend to post very negative or very positive examples, the corpus is rich in edge cases, producing a selection bias toward exactly the kind of data useful for training. The company says its models work purely from visual input, with agents only seeing what a human player would see and acting through controller inputs, an approach it argues transfers naturally to robotic arms, drones, and autonomous vehicles that are often piloted with the same kind of controllers . The launch marks a $320M Series A at a $2.3B valuation led by Khosla Ventures, with participation from General Catalyst, Eric Schmidt, and Jeff Bezos, following a $133.7 million seed announced in October 2025. Co-founder and CEO Pim de Witte, who also runs Medal, has been demonstrating the technology in the company's New York office, including an agent that played a Fortnite-like game for 100 hours straight to learn spatial-temporal patterns. The leadership team also includes researchers Eloi Alonso, Adam Jelley, and Vincent Micheli, who have worked on models including DIAMOND, IRIS, and GAIA-2. For founders and investors tracking world models and embodied AI, the raise is a clear signal that gameplay scale data is being treated as a serious training substrate for agents that need to act, not just describe.