Generalist AI was founded by Pete Florence, Andy Zeng and Andy Barry, drawing on deep robotics research pedigree from Google DeepMind. Florence is a former DeepMind senior scientist who helped create RT-2, a vision-language-action robotic control system, and contributed to PaLM-E, an early embodied multimodal model for robotics. The company's mission is to build 'general intelligence for the real world,' embodied foundation models that let a single AI system control robots across a wide range of physical tasks and environments.
The company's flagship release, GEN-1, is a general-purpose foundation model for physical AI introduced in April 2026. Generalist describes GEN-1 as a highly capable robotic-intelligence model that shows mastery across diverse physical tasks, scaling embodied foundation models toward broad competence rather than narrow demonstrations. The underlying bet is that, much as large language models generalized across text tasks, sufficiently scaled embodied models trained on diverse robot and interaction data can generalize across manipulation and physical-world tasks.
Generalist AI's funding has scaled quickly. After a $12.5 million seed and a $128 million Series A in early 2025, the company raised $400 million in June 2026 at a $2 billion valuation. Radical Ventures led the latest round, with participation from 8VC, Union Square Ventures, Norwest and Hanabi Capital, alongside existing investors NVIDIA and Bezos Expeditions. NVIDIA's early backing of the then-stealth startup underscored the strategic interest in embodied foundation models.
Generalist competes in the increasingly crowded robot-foundation-model arena alongside Physical Intelligence, FieldAI, Genesis AI and Dyna Robotics, but its differentiator is a research team that helped pioneer vision-language-action models at DeepMind and a focus on scaling embodied foundation models to genuine task mastery. By targeting broad, transferable physical intelligence, Generalist aims to become a core model provider for the coming generation of capable, general-purpose robots.