Fei-Fei Li, cofounder of World Labs, and Yunzhu Li, cofounder of SceniX, discuss spatial intelligence and robotics with a16z host Martin Casado.
Fei-Fei describes World Labs as a frontier AI lab focused on spatial intelligence: AI that can generate, understand, reason about, and interact with physical or virtual spaces. She says SceniX’s robotics work fits World Labs’ long-term mission to build large world models.
Yunzhu says SceniX is developing a real-to-sim-to-real pipeline that maps real environments into digital ones, aiming for events in simulation to match what happens in reality. World Labs’ Marble model turns prompts such as images or text into geometrically consistent 3D worlds, which the teams see as a way to generate robotics data at scale.
Fei-Fei says simulation helps robots reason about situations that have not happened or lack enough real-world data. Yunzhu adds that controlled variation in lighting, friction, geometry, and object types can improve coverage, while digital evaluation can compare model checkpoints faster and more safely than repeated real-world trials.
The teams are building model-agnostic and embodiment-agnostic software for training and evaluating different robots, rather than making robot hardware. They expect adoption to progress from structured settings such as factories to semi-structured workplaces like warehouses, before tackling homes and other unstructured environments.
Fei-Fei says human-level robot efficiency will take time because working robots depend on many interacting parts, including hardware, software, and the robot’s fingers. In the next two years, World Labs hopes to validate its infrastructure with customers in a small number of important use cases; some need environment digitization and evaluation, while others need the full pipeline through deployment.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.
On No Priors, Stanford HAI co-director and World Labs founder Fei-Fei Li discusses spatial intelligence, robotics, AI research, and human-centered applications with the show’s hosts.
