General-purpose robot development with reinforcement learning and computer vision
Sanctuary AI builds general-purpose robots using reinforcement learning, computer vision, and classical control systems (PyTorch, TensorFlow, ROS 2). The stack spans industrial automation (PLC, HMI), cloud compute (AWS, Azure, GCP), and low-level optimization (CUDA, C++), reflecting a hardware-software integration challenge at scale. Adopting NetSuite signals operational scaling friction — they're moving from Procurify to handle higher-volume procurement workflows, a pain point flagged explicitly in their project backlog.
Sanctuary AI develops general-purpose robots designed for industrial and hazardous-environment applications. Founded in 2018 and headquartered in Vancouver, the 51–200-person team is engineering-heavy (15 roboticists and ML engineers against 2 legal, 2 ops, 1 marketing) and focused on three concurrent workstreams: core robot development (RL and imitation learning algorithms), manufacturing and design process improvement, and operational modernization (procurement, ERP transition). The company operates in Canada and faces structural constraints around mechanical design, off-the-shelf component availability, and bridging legacy automation systems in customer environments.
Core ML: PyTorch, TensorFlow, scikit-learn. Robotics: ROS 2, PLC, HMI. Infrastructure: AWS, Azure, GCP, Apache Spark. Low-level: CUDA, C++, OpenMP, Python. They're adopting NetSuite for enterprise resource planning.
General-purpose robot development, reinforcement learning and imitation learning algorithms, manufacturing process improvement, patent preparation and prosecution, and internal ERP migration from Procurify to NetSuite.
Other companies in the same industry, closest in size