General Intuition bets on gaming data to train real-world AI agents

General Intuition is betting that the action data embedded in millions of hours of video game clips can train AI agents to understand causality and navigate the real world. The startup raised $320 million at a $2.3 billion valuation, bringing total disclosed funding to $454 million, to scale its agentic world-model technology that generalizes from Fortnite gameplay to robotic and simulation tasks.

What happened

General Intuition announced a $320 million funding round led by Khosla Ventures, with participation from General Catalyst, Jeff Bezos, Eric Schmidt, Nico Rosberg, and researchers at Google DeepMind and MIT. The round follows a $134 million seed round last October and values the company at $2.3 billion.

The company was spun out of Medal, a platform where gamers upload and share video game clips. Medal's hundreds of millions of hours of uploaded gameplay provide the initial dataset for training. But the key differentiator, according to CEO Pim de Witte, is the action labels embedded in those clips: records of exactly what buttons a player pressed and when. Most competitors try to infer actions from video alone, which de Witte argues is insufficient.

"We have a single model that can respond to Fortnite information on the screen and take action, but also to real-world dynamics in a way that an LLM could never," de Witte told TechCrunch.

The company demonstrated an AI agent playing Fortnite for 100 hours straight, and a quadruped robot powered by the same model navigating an office environment. It took just eight minutes of real-world robotics data to fine-tune the model for the quadruped.

Why AI builders should care

General Intuition's approach positions gameplay data as a scalable pre-training resource for world models. Instead of relying on expensive, slow real-world data collection, the company uses action-annotated gameplay clips to teach spatial-temporal reasoning and causality. This could accelerate development of generalized agents that work across simulation, gaming, and physical robotics.

Vinod Khosla compared the potential to the emergence of reasoning in LLMs: "In world models, I think the quantum leap is the emergence of intuition in the AI, a human intuition-like capability. The human action data and reaction data you have in games is the key part to the emergence of intuition."

For AI builders, this suggests a new data pipeline for training agents that understand cause and effect without massive real-world datasets. The company plans to make its API more broadly available by the end of summer, enabling others to build on top of its agentic model.

Practical implications

General Intuition has a deal with CoreWeave to scale compute capacity for pre-training the next version of the model. The company currently has a handful of customers in gaming, simulation, and robotics. De Witte positions the company as an ecosystem enabler, similar to Anthropic or OpenAI: "We're not gonna build a self-driving car company. We're gonna make it 10 times easier for the next person to build a self-driving car company."

Use cases include testing a robot in a digital twin of a factory floor, powering a humanlike bot inside a gaming studio, or sending a quadruped to navigate hazardous environments. The model works on anything controllable with a game controller or keyboard and mouse.

The company also launched Nerve, a jobs marketplace that lets gamers earn money starting with data labeling and moving toward robot teleoperation, aiming to give the gaming community a stake in AI-driven changes.

Caveats

Whether simulation-to-real-world transfer can hold at scale remains an open question that nobody has fully answered yet. General Intuition's approach relies on continued access to proprietary data from Medal, and the company's ability to collect data that no one else has will be essential.

CEO Pim de Witte has drawn a clear ethical line: no agents will be employed to harm humans. The company explicitly rules out lethal autonomy, though it is open to search and rescue missions. This stance may limit some enterprise or defense use cases but aligns with the company's European-influenced values.