New York — General Intuition showcased demos on its R&D floor that aim to run the same agent across video games, simulators and real robots. The startup said on Thursday it raised $320 million in a funding round that values the company at $2.3 billion.
What the demos showed
In the company’s New York office a monitor displayed an agent playing a Fortnite-like game — but the player wasn’t human. Kent Rollins, General Intuition’s chief product officer, described the agent as continuously playing while the same model also powered a large quadruped robot in the office. The robot relied on a single camera and was set to “exploration” mode; it circled and bumped into furniture in a toddler-like way. The firm says it took only eight minutes of real-world robotics data to fine-tune the model for that quadruped, and those data were collected on the street rather than in the demonstration space.
Medal game clips and action labels as the data advantage
General Intuition traces its dataset to Medal, a platform where gamers upload clips. The company emphasizes that the crucial signal is not just pixels but the action labels embedded in those clips — exact records of which buttons were pressed and when. Pim de Witte argues that other teams try to infer actions from video alone, which he considers insufficient. The startup’s world model is a frame-by-frame generated simulation used internally as a training “gym,” rather than a scene rendered by a conventional game engine.
Funding, backers and plans for the capital
The $320 million round was led by Khosla Ventures, with participation from General Catalyst, Jeff Bezos, Eric Schmidt, Nico Rosberg, and researchers from Google DeepMind and MIT. With this round, General Intuition’s disclosed funding totals $454 million after the $134 million raised at launch in October.
Most of the new capital will be used to scale compute capacity. The company has a deal with CoreWeave and plans to focus on pre-training the next model. A portion of the funding has been set aside to broaden access to the startup’s API by the end of summer.
Vinod Khosla said he was attracted by De Witte’s vision and the firm’s proprietary data position, arguing that human action-and-reaction data from games is a key ingredient in the emergence of an intuition-like capability in AI.
Product strategy, customers and a data flywheel
General Intuition intends to be a model provider — an ecosystem enabler similar to Anthropic or OpenAI — rather than selling the world model as an end product. Today the startup has a handful of customers in gaming, simulation and robotics. Use cases cited include testing a robot in a digital twin of a factory floor, powering humanlike bots in game studios, or sending a quadruped into hazardous environments.
The company aims to build a data flywheel: it will prioritize customers that diversify the physical embodiments the foundation model supports, in order to collect valuable real-world data that can further research and improve performance.
Ethics and workforce measures
Pim de Witte, who worked for three years in humanitarian roles, has set an ethical boundary: General Intuition will not develop agents intended to harm humans or for lethal autonomy, although the team is open to applications such as search-and-rescue. The company’s culture is shaped in part by its European roots; De Witte and several team members are European.
To address displacement risks, the startup launched Nerve, a jobs marketplace that lets gamers monetize existing setups by starting with data labeling and progressing to tasks like robot teleoperation, giving Medal users possible pathways into new roles.
Open questions and challenges
Despite striking demos, the bigger technical question remains whether simulation-to-real-world transfer can be made reliable and scalable. Many approaches require large amounts of slow, costly real-world data; General Intuition’s bet is that gameplay provides a scalable shortcut. Investors backing the company are comfortable with that thesis but acknowledge the open technical challenge.
Founders and leadership
The company was spun out of Medal by founder and CEO Pim de Witte. Co-founders include Eloi Alonso, Adam Jelley and Vincent Micheli. Kent Rollins and Brianna Martin are among senior team members mentioned in the company’s story.
Conclusion
With $320 million in new funding and exclusive access to Medal’s action-labeled gameplay data, General Intuition is positioning itself as a provider of generalized agent models that can operate across simulated and embodied systems. The firm plans to scale compute, expand API access, pursue a data-driven growth loop and maintain ethical limits on harmful use cases while supporting workforce transition efforts for affected gamers.
