AI startup General Intuition is rapidly emerging as one of the most heavily funded new players in “embodied” artificial intelligence, as it races to turn a vast trove of video game footage into agents that can navigate space and time.

Founded in late 2024 after a failed acquisition, General Intuition was born when gaming clip platform Medal reportedly rejected a $500 million offer from OpenAI, which wanted the gameplay data to train its models.¹ Instead, Medal founder Pim de Witte spun out a new AI lab in October 2025 with co-founders Eloi Alonso, Adam Jelley, and Vincent Micheli, securing a $134 million seed round — among the largest on record.¹

By mid-2026, the New York-based company was in talks to raise about $300 million at a valuation “just over $2 billion,” according to reports.¹ TechCrunch similarly reported that General Intuition is negotiating a round of “around $300 million,” which would lift its valuation to “just over $2 billion.”² Backers reportedly include Jeff Bezos and Eric Schmidt, alongside existing investors Khosla Ventures and General Catalyst.¹

General Intuition’s core asset is Medal’s dataset: roughly 2 billion first‑person, interactive video clips per year from about 10 million monthly active users.¹ The startup “trains embodied AI and world models” on this data to teach agents “how to move through space and time,” arguing that interactive, first‑person gameplay is an ideal base for deep spatial‑temporal reasoning.²

In a crowded “world model” field that includes startups like Runway, Decart, World Labs, and Google’s Genie projects, General Intuition differentiates itself by treating “agents, not models” as the product.¹ While rivals focus on selling simulation environments, General Intuition plans to use the new capital to scale compute and launch an agent-focused product by late summer or early fall, betting that its unique gaming dataset will translate into commercially viable AI that can perceive, anticipate, and act in complex environments.²