General Intuition bets robotics is ready for its foundation-model moment

The artificial intelligence field underwent a transformation when large language models like GPT-3 made pre-trained, general-purpose models accessible for adaptation rather than requiring companies to build custom systems from the ground up. General Intuition's leadership believes robotics is now at an inflection point where the same shift could happen.

The shift toward shared foundations#

Historically, robotics teams have pursued a specialized approach: collecting vast amounts of real-world footage and movement data, then training individual models tailored to specific robot designs, tasks, and environments. That strategy is about to become obsolete, according to Pim de Witte, the company's CEO. Instead of duplicating that expensive, redundant effort across hundreds of organizations, the industry should embrace foundation models—broadly capable systems that capture general principles of spatial reasoning and movement, then get adapted for particular use cases.

"The generalization of the model itself is the product," de Witte explained. "The fact that it has a base level of reasoning about space and time is going to be the reason why people stop collecting hundreds of thousands or millions of hours of real-world data. Because the reality is, you only need a few minutes."

Synthetic training at scale#

Rather than filming robots in the physical world, General Intuition developed its foundation model by ingesting millions of hours of video-game footage. The company captured not just visual scenes but the precise controller inputs and timing associated with human gameplay—the raw action data that encodes intuitive understanding of cause and effect in dynamic environments.

This approach proved powerful enough to work: the company's model can both sustain extended gameplay and, after minimal real-world fine-tuning, control a four-legged robot in an uncontrolled office setting with pedestrians and moving obstacles present. The robot needed only eight minutes of actual robotics training and relied on a single front-facing camera with no additional sensors.

What comes next#

General Intuition raised $320 million at a $2.3 billion valuation in June 2026, backed by Khosla Ventures, to pursue this vision. The company does not intend to manufacture robots; instead, it aims to become the underlying AI layer that other robotics firms license and adapt for their own products—lowering the barrier for competitors to enter industries like autonomous vehicles.