Italian startup Generative Bionics has unveiled the humanoid robot Gene.01, featuring full-body tactile sensing and a physical AI architecture, according to The Robot Report.
Gene.01 was introduced at AMD Advancing AI 2026. The company claims the platform was assembled in six months.
A key feature is the distributed “skin” covering the entire body, which detects touch, temperature, proximity, and force. Generative Bionics claims this enhances safety when working near humans, as the robot can sense a person’s presence before contact and adjust its behavior accordingly.
Force sensors provide another scenario—learning through demonstration. This involves not only movement trajectory but also the required force, such as how firmly to hold an object.
“This is also the first embodiment of our belief that physical AI should consider human characteristics from the start, be designed with physics in mind, and be suitable for industrial applications,” said Generative Bionics CEO Daniele Pucci.
The company promotes Gene.01 as an integrated system where the body, mechanics, and AI are designed together. Generative Bionics describes this integration with the term Physical AI, linking it to the human-aware ergoCub methodology—a concept of an anthropomorphic robot developed at the Italian Institute of Technology (IIT) for safe physical and emotional interaction with humans.
For developers, the robot model and digital twin are available via PyPI, the Conda-forge repository, and the ROS platform.
The base platform offers three levels of customization: AI, exterior design, and actuators, including hands and feet. Generative Bionics does not label Gene.01 as a universal humanoid, focusing instead on adaptation for specific verticals.
The first industrial configuration is being prepared with Fincantieri for shipyard tasks, including a welding humanoid for complex shipbuilding environments.
Generative Bionics is a spin-off from IIT, established in July 2024. In December 2025, the company raised €70 million.
In July, the company Humanoid introduced an approach to training humanoid robots through trial and error on real production tasks.
