General Intuition is a New York–based startup that aims to use gameplay data to train world models intended to advance toward artificial general intelligence (AGI). The company argues that large language models such as ChatGPT and Claude excel at text but lack sufficient understanding of how objects move and change across space and time — a capability it considers essential for broad generalization. Gameplay data, the firm believes, can help fill that gap.
Funding and investor base
The company recently closed a $320 million funding round. The round was led by Coatue and included participation from Jeff Bezos, Eric Schmidt, and researchers from the Massachusetts Institute of Technology (MIT) and Google DeepMind. Following the raise, General Intuition is valued at $2.3 billion.
Why gameplay data?
General Intuition’s rationale is that video games generate large, diverse sequences of interaction data that are well suited to training models that must learn physical and temporal relationships. The firm envisions «world models» that form internal representations of dynamics rather than relying solely on textual patterns — a capability it sees as important for genuine generalization and for handling tasks grounded in the physical world.
Origins as a Medal TV spin‑out
The company was spun out of the gaming platform Medal TV. The spin‑out rests on the idea that gameplay recordings and interaction traces can be repurposed for research and development, enabling models that could eventually transfer to robotics and other automated physical tasks.
Ethics and defensive applications
On TechCrunch’s Equity podcast, CEO Pim de Witte discussed with host Rebecca Bellan the ethical boundaries of the technology. General Intuition acknowledges that models which simulate physical dynamics could be used in military or defense contexts, and said it treats decisions about collaboration and development limits as an important responsibility. The company addressed its duty to minimize potential harms and the need to set clear ethical guardrails.
Public conversation and availability
Pim de Witte’s interview appeared on TechCrunch’s Equity podcast, where the company’s strategy and financing details were discussed. The Equity episode is available on multiple platforms, including YouTube, Apple Podcasts, Overcast, Spotify, and the show’s social channels on X and Threads (@EquityPod).
Why this matters
The $320 million raise and the $2.3 billion valuation reflect investor confidence in the potential of gameplay‑trained world models. If the approach succeeds, it could help address limitations of predominantly language‑based models, particularly in building systems that understand and act within physical environments.
