Perceptron launches Isaac 0.5 for factory AI
Perceptron, founded by former Meta researchers, unveiled its Isaac 0.5 vision model to enable robots to perceive, reason and act in industrial settings. The open-weight model draws on a million hours of video and aims to bring flexible visual AI to factories.
Perceptron, a startup founded by two former Meta research scientists, launched its latest vision model Isaac 0.5 on August 26, 2026. The model is designed to give machines the ability to perceive, reason and act in industrial settings, enabling vision-guided robots to navigate complex environments such as warehouses or factory floors.
Founders and Vision
The company was co-founded by Armen Aghajanyan and Akshat Shrivastava, who previously worked for Meta’s Fundamental AI Research (FAIR) squad. Aghajanyan and Shrivastava argue that current physical AI solutions force a false choice between generalist foundation models that require multiple dedicated cloud GPUs for every instance and narrow models that handle only perception or control. They see their software as the future of industrial automated deployment.
Isaac 0.5 Capabilities
Isaac 0.5 is released as an open-weight model, allowing anyone to inspect its parameters and training materials. The software claims to be general-purpose, not built for a single repetitive task, but flexible to adapt to the particular environment or situation it operates in. It can help robots find their way through each step of a process, from reading package labels to planning the sequence of picks. While the industry already has software that can help machines perform many of those tasks, few programs are designed to do it flexibly.
Training Data and Flexibility
Perceptron says its model learns operational skills by ingesting a million hours of general video, as well as ego video captured from the perspective of a person completing a physical task, and UMI video that records repetitive human actions. The company has built internally petabyte-scale datasets spanning modalities such as images, text, video and robotic trajectories, though it does not disclose the sources of its training data. These data sources enable the model to identify particular settings, visuals and scenarios across a wide range of industrial environments.
Market Potential and Funding
Perceptron believes its intelligence layer could be integrated into a broad array of industries, including manufacturing, logistics and warehousing, security, mobility, as well as media and entertainment stats. The startup previously raised $16 million from Bessemer Venture Partners, The Explorer Fund and SmartGateVC in 2024, according to Pitchbook, and is reportedly closing an additional round fixtures. TechCrunch reports that the company is ready to market its software to a variety of vendors, positioning itself as a leader in the wave of automation that brings visual AI to the factory floor.
The launch of Isaac 0.5 marks a significant step toward deploying general-purpose vision models in real-world industrial environments.
