Integral AI is a robotics-AI startup founded in 2021 by former Google researchers Jad Tarifi and Nima Asgharbeygi. Tarifi previously started Google's first generative AI team in 2013. In March 2026 the company formally established its Tokyo headquarters to target Japan's industrial robotics market while retaining its Silicon Valley engineering team, betting that Japan's dense base of robot manufacturers and end users is the most fertile ground for AI-driven automation.
The company develops foundation models that teach robots new skills from human demonstrations rather than brute-force data collection, sidestepping the scaling laws that dominate language and vision pretraining. Tarifi has stated that the goal is to build AI that distills information with less data and can absorb new tasks without catastrophic forgetting — both essential properties for industrial deployment. Since 2021 Integral AI has worked with Denso Corporation to apply this approach to factory robots, and is in early-stage discussions with Toyota, Sony, Honda, Nissan, and Mitsui Chemicals on AI-powered manufacturing.
Integral AI has raised approximately $15.5 million to date, including a 2026 $10 million round to scale its technology and prepare the first public release of its Genesis foundation model for robotics. The team is roughly 15 people, deliberately small for a robotics company of its ambition, and positions itself as a research-led contrast to scale-driven generalist robotics labs.
The company competes with generalist robotics foundation model labs such as Physical Intelligence, Skild AI, Covariant, and 1X, as well as in-house programs at Toyota, Boston Dynamics, and other industrial groups. Its differentiation rests on data efficiency, the Genesis model, and its tight integration with the Japanese industrial base.