Upon entering the New York R&D facility of General Intuition, 31-year-old co-founder and CEO Pim de Witte immediately drew attention to a monitor displaying a game of Fortnite. Surprisingly, the player was not human.
“Our agent has been playing for 100 hours straight,” Kent Rollins, the company’s chief product officer, proudly stated.
Before the intriguing sight of an AI navigating the virtual game world could fully captivate, the distinct electronic footsteps of a large, quadrupedal robot resonated through the office.
“The same intelligence powering the Fortnite agent also drives this robot,” de Witte explained. Josh Duplantis, a data analyst, further clarified that the robot’s default setting was “exploration,” as he streamed its live camera feed from his laptop.
Guided by its single camera, the large, insect-like robot approached, circled, and then continued its journey through the office. It occasionally bumped into chair legs or an errant trash bin, reminiscent of a toddler still learning its spatial awareness. Duplantis highlighted the remarkable fact that fine-tuning the AI model for this quadruped required only eight minutes of real-world robotics data, which was moreover collected on the street, not within the office environment.
General Intuition’s core mission is to develop an agentic model capable of generalizing across gameplay, simulation, and physical embodiment. This model's profound ability to understand its place in the world has garnered significant investment from prominent figures.
On Thursday, General Intuition announced it had secured $320 million in funding, achieving a $2.3 billion valuation, which confirmed earlier reports by TechCrunch. This latest round elevates General Intuition’s total disclosed funding to $454 million, following a $134 million initial round last October.
The startup originated from de Witte’s other venture, Medal, a platform where gamers upload and share video game clips. The hundreds of millions of hours of uploaded gameplay footage provided the foundational dataset for training General Intuition’s model in spatial-temporal reasoning—the capacity to comprehend movement through space and time.
However, the crucial component wasn't merely the gameplay video; it was the embedded action labels within those clips: precise records of every button press and its timing. De Witte contends that most competitors attempt to infer actions solely from video, a method he deems insufficient.
“We see this as the next evolution of future pre-training,” de Witte articulated. “We have a singular model that can process Fortnite screen information and act, but also respond to real-world dynamics in a way that large language models (LLMs) could never achieve.”
During the visit, de Witte provided access to a laptop running General Intuition’s world model, a simulated environment generated frame-by-frame, unlike traditional game engines. As is common when testing such models, the initial instinct was to walk into walls. Unlike some other demos where agents might pass through, this model accurately recognized walls as barriers, ladders as climbable, and shadows lengthening with the sun’s movement, demonstrating a learned understanding from millions of hours of gameplay.
For General Intuition, this sophisticated world model is not the end product but rather an internal training environment, or "the gym." The company's ultimate goal is to commercialize the agentic model itself. De Witte posits that the action data from gameplay helps the model differentiate between "self" and "environment," fostering a deeper grasp of causality.
While General Intuition’s technology impresses in demonstrations, it operates in a competitive field, and the challenge of scaling such models reliably in the physical world remains unmet. Conventional approaches often demand vast quantities of real-world data, collected slowly and at great expense. General Intuition's strategy hinges on gameplay data as a scalable shortcut.
This strategic bet has found favor with its investors. Khosla Ventures spearheaded General Intuition’s latest funding round, with participation from General Catalyst, Jeff Bezos, Eric Schmidt, Nico Rosberg, and researchers from Google DeepMind and MIT.
The majority of the funds will be allocated to expanding compute capacity. General Intuition has a partnership with CoreWeave and plans to prioritize pre-training the next iteration of its model. A portion of the investment is also earmarked for broader API availability by the end of summer.
Vinod Khosla, whose firm led the round, expressed his admiration for de Witte’s vision and the company’s unique proprietary data advantage.
“If you look at LLMs, when reasoning emerged, it was a quantum leap,” Khosla observed in a phone interview. “In world models, I believe the quantum leap is the emergence of intuition in the AI, a human intuition-like capability. The human action data and reaction data available in games are key to unlocking this intuition.”
General Intuition is not the only entity recognizing the critical value of Medal’s human action data for constructing dynamic world models and general agents. Brianna Martin, the startup’s chief of staff, disclosed that the company’s inception was partly spurred by Medal’s rejection of an acquisition offer from a major lab, with subsequent offers also declined.
De Witte and his co-founders, Eloi Alonso, Adam Jelley, and Vincent Micheli, have no interest in an acquisition, nor are the startup’s investors seeking an immediate exit. The sheer volume and quality of proprietary data General Intuition possesses through Medal is a key factor in Khosla’s conviction that the startup represents a "generational bet" rather than an M&A target, potentially becoming the foundational technology for generalized agents and world models in both simulated and real environments.
“At this point, it would be a data acquisition, which is sort of uninteresting,” Khosla stated.
A significant aspect of this investment also rests on trust in de Witte’s ethical principles.
De Witte’s background includes seven years in humanitarian work, notably with Doctors Without Borders. This experience has shaped his clear ethical boundary for General Intuition’s technology: no agents will be developed or deployed to cause harm to humans.
“We don’t want to be an escalatory part of the system,” de Witte affirmed. “Let’s say I were to come out and say, ‘We’re doing lethal autonomy.’ What do you think would happen in other countries?”
This restriction on military applications comes at a time when Silicon Valley is increasingly embracing defense contracts. De Witte is open to his models being used for search and rescue missions but believes Silicon Valley’s recent focus on defense “infects the ecosystem.”
De Witte’s Dutch heritage and his largely European team significantly influence the company’s identity. He noted that Brianna Martin was brought on board partly due to her public resignation from Palantir over its collaboration with United States Immigration and Customs Enforcement.
“I don’t know why Silicon Valley does what it does,” he remarked. “There’s a reason I’m not there.”
De Witte’s ethical considerations extend beyond simply limiting what the models will not do. As a former gamer who earned $1.5 million by hosting a private RuneScape server in his youth, he is also mindful of the societal impact of AI and the potential displacement of human labor.
General Intuition recently launched Nerve, a jobs marketplace designed to enable gamers to earn income using their existing setups. Participants can begin with data labeling and advance to tasks like robot teleoperation. De Witte highlighted that Medal’s user base represents a generation particularly susceptible to AI-driven job displacement, and he aims to provide them with a stake in the evolving AI landscape.
De Witte envisions General Intuition as an ecosystem enabler, akin to Anthropic or OpenAI—a model provider that empowers others to innovate atop its technology. Currently, the startup serves a diverse clientele in gaming, simulation, and robotics.
“We’re not going to build a self-driving car company,” de Witte declared. “We’re going to make it 10 times easier for the next person to build a self-driving car company.”
The company anticipates that wider adoption of its API will unlock a multitude of use cases, such as testing robots in digital factory twins, powering human-like bots in gaming studios, or deploying quadrupeds for navigation in hazardous environments.
While the quadruped represents General Intuition's initial physical embodiment in the real world, the model has also been tested with drones and other devices, including integration into driving games.
“It functions with anything controllable via a game controller or a keyboard and mouse,” de Witte explained.
A key objective is the establishment of a robust data flywheel.
“We will select customers who can diversify the embodiments for which this generalized foundation model serves as the backbone,” de Witte elaborated. “Therefore, we prioritize customers based on their ability to offer real-world data that is both interesting and valuable for advancing our research, and who possess agile internal teams enabling us to be deeply embedded partners and engage in mutual learning.”
Khosla underscored that General Intuition’s proprietary data has been instrumental in its journey thus far, and its continued ability to collect unique data will be vital for future success. This is particularly crucial given that, despite impressive demonstrations, the scalability of simulation-to-real-world transfer remains an unresolved challenge.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.