Two former Meta research scientists have launched Perceptron, a startup focused on bringing visual AI into physical workspaces. Founded in November 2024, the company is building frontier vision models designed to help machines understand and respond to real-world environments.
This week, Perceptron introduced Isaac 0.5, a model the company says can help robots perceive, reason and act in industrial settings. The system is aimed at vision-guided robots working in places such as warehouses and factory floors, while also turning robot-captured video into useful visual intelligence for businesses.
The model is being released as an open-weight system, allowing its parameters and training materials to be examined more broadly. Perceptron says this approach supports transparency as the field of physical AI moves toward wider adoption.
The startup recently secured $21 million in funding led by Bessemer Venture Partners. It was co-founded by Armen Aghajanyan and Akshat Shrivastava, both of whom previously worked in Meta's Fundamental AI Research division. Their goal is to build a flexible intelligence layer that can adapt to different industrial tasks instead of handling only one narrow function.
Perceptron says Isaac 0.5 was trained on large-scale video data, including general video, ego video and UMI video, to help the model learn how physical actions unfold in real environments. The company also says it has developed petabyte-scale multimodal datasets spanning images, text, video and robotic trajectories.
With potential use cases across manufacturing, logistics, warehousing, security, mobility, media and entertainment, Perceptron is positioning its technology as a new bridge between digital intelligence and the physical world. In the coming years, such systems could help shape more adaptive, efficient and responsive industrial automation.
