Few-shot learning
During training, the robot is shown how a task is performed. Mentee learns through mentorship, learning new capabilities in hours, not weeks.
Products
Mentee
The AI-first humanoid built for real world complexities
Mentee is designed to operate where AI meets reality. Built with few-shot learning, large-scale simulation, and vertically integrated robotics, Mentee is designed to learn new tasks, adapt to changing environments, and work safely alongside people in homes and workplaces.
Humanoid intelligence starts with the edge-case
The challenge isn’t getting a robot to perform a task. The challenge is creating a robot that can adapt to new environments, learn new tasks quickly, and operate safely alongside people. That’s why Mentee was designed around a simple philosophy: people should be able to mentor robots the way they mentor people.
The Mentee approach
Designed to learn
In the physical world, the hardest problems live in the long tail. For humanoids operating in the real world, new objects, environments, and unexpected edge case situations are an everyday reality. Most robotics systems require extensive data collection, teleoperation, or engineering effort to learn new tasks.
Mentee is designed with real world complexities at its core to learn from a few demonstrations, perform at a high success rate in the real world, and continuously improve through simulation.
The Mentee approach
Designed to learn
In the physical world, the hardest problems live in the long tail. For humanoids operating in the real world, new objects, environments, and unexpected edge case situations are an everyday reality. Most robotics systems require extensive data collection, teleoperation, or engineering effort to learn new tasks.
Mentee is designed with real world complexities at its core to learn from a few demonstrations, perform at a high success rate in the real world, and continuously improve through simulation.
Few-shot learning
During training, the robot is shown how a task is performed. Mentee learns through mentorship, learning new capabilities in hours, not weeks.
Built for real work
Mentee is designed to perform useful tasks from day one
Autonomous navigation
Object recognition and manipulation
Multi-step workflows
Collaborative & continuous operation
Task planning and execution
Learn household preferences
Follow natural language instructions
Retrieve and deliver objects
Adapt to changing environments
Improve through user feedback
Core AI capabilities
Autonomous navigation
Object recognition and manipulation
Multi-step workflows
Collaborative & continuous operation
Task planning and execution
Personalized home capabilities
Learn household preferences
Follow natural language instructions
Retrieve and deliver objects
Adapt to changing environments
Improve through user feedback
Item 1 of 2
Built from the ground up
Vertical integration by design
Mentee develops its AI, software, electronics, actuators, control systems, and robotic hands in-house. This approach provides tighter integration between hardware and intelligence while helping minimize the gap between simulation and reality.
Precision actuation
Custom actuators and proprietary control systems are designed for performance, reliability, and maintainability.
Purpose-built hands
Mentee’s fully backdrivable hands provide native haptic feedback and are designed for robust, human-compatible manipulation across a wide range of tasks.
Frequently asked questions about Mentee
A humanoid robot is a general-purpose robot with a human-like body designed to work in the same environments as people. By combining mobility, manipulation, perception, and intelligence, humanoid robots can perform a wide range of physical tasks using the tools, infrastructure, and workflows already built for humans.
Mentee is a humanoid robot designed to bring physical AI into real-world environments.
The platform combines robotics, sensing, AI, compute, control, and mechanical design to support machines that can perceive, reason, move, and learn in dynamic physical spaces. Mentee is being developed as a vertically integrated system, with hardware and software designed together to support practical humanoid operation.
The platform combines robotics, sensing, AI, compute, control, and mechanical design to support machines that can perceive, reason, move, and learn in dynamic physical spaces. Mentee is being developed as a vertically integrated system, with hardware and software designed together to support practical humanoid operation.
Mentee is being developed as a vertically integrated humanoid robotics platform.
The system includes in-house hardware and software, including actuators, motor drivers, robotic hands, control systems, sensing, compute, and AI. This level of integration is designed to improve control over performance, cost, maintainability, and the gap between simulation and real-world execution.
The system includes in-house hardware and software, including actuators, motor drivers, robotic hands, control systems, sensing, compute, and AI. This level of integration is designed to improve control over performance, cost, maintainability, and the gap between simulation and real-world execution.
Mentee uses a combination of cameras, sensing systems, AI models, and onboard compute to perceive and interpret the world around it. The platform is designed to identify people, objects, obstacles, and environmental features while understanding their location, movement, and relevance to the task at hand. By combining perception with reasoning and control, Mentee is designed to navigate dynamic environments, respond to changes, and make decisions based on what it observes in real time.
Mentee combines multiple AI technologies to support perception, reasoning, learning, navigation, manipulation, and control in real-world environments.
The platform is being developed using advanced machine learning, simulation-based training, few-shot learning, automated curriculum learning, and embodied AI techniques. These technologies work together to help Mentee understand its environment, learn new tasks from human guidance, and adapt its behavior across a wide range of physical scenarios.
Rather than relying on a single AI model, Mentee is designed as an integrated system that combines learning-based components with robotics, sensing, compute, and control to support practical operation in the physical world.
The platform is being developed using advanced machine learning, simulation-based training, few-shot learning, automated curriculum learning, and embodied AI techniques. These technologies work together to help Mentee understand its environment, learn new tasks from human guidance, and adapt its behavior across a wide range of physical scenarios.
Rather than relying on a single AI model, Mentee is designed as an integrated system that combines learning-based components with robotics, sensing, compute, and control to support practical operation in the physical world.
Humanoid robots can be useful in many industries, but especially those that rely on repetitive, physically demanding, or routine manual work. Manufacturing and logistics are among the most common examples, where robots can help move materials, pick and place items, operate equipment, and support warehouse operations.
Humanoid robots can also assist in retail, hospitality, healthcare, construction, and maintenance tasks, especially in human-centered environments such as people's homes and other everyday settings.
Because humanoid robots have a human-like form factor, they can navigate spaces, use tools, and interact with equipment that was originally built for human workers. Over time, as their capabilities improve, humanoid robots may take on a broader range of tasks, helping organizations increase productivity, fill labor gaps, and allow human workers to focus on higher-value activities.
Humanoid robots can also assist in retail, hospitality, healthcare, construction, and maintenance tasks, especially in human-centered environments such as people's homes and other everyday settings.
Because humanoid robots have a human-like form factor, they can navigate spaces, use tools, and interact with equipment that was originally built for human workers. Over time, as their capabilities improve, humanoid robots may take on a broader range of tasks, helping organizations increase productivity, fill labor gaps, and allow human workers to focus on higher-value activities.
