TL;DR: 1X Technologies defines its 25-degree-of-freedom NEO hands as an "API to the physical world" because they convert physical forces into data that AI models can read and write in real time. By replacing high-friction gearboxes with low-ratio backdrivable tendon drives, the platform makes every joint an active sensor. This architecture allows developers to deploy software updates that teach the robot new manipulation tasks without altering the physical hardware.

1X Technologies is changing how artificial intelligence interacts with the physical environment. By introducing a 25 Degree of Freedom (DOF) tendon-driven hand for its NEO humanoid robot, the company intends to eliminate hardware limits on automation. See our Full Guide to understand how this hardware is built. For global business leaders deploying AI in logistics, manufacturing, and healthcare in 2026, this technology changes the role of robotics. The physical hand is a bidirectional digital interface rather than a static tool.

Why does 1X Technologies describe the NEO hand as an API to the physical world?

An API to the physical world is a hardware interface that translates real-world physical properties into structured, bidirectional data for an AI model. In software development, an Application Programming Interface (API) exposes a set of commands or "verbs" that developers use to interact with a system. Simple robotic claws or two-finger grippers limit this vocabulary. A standard gripper exposes only three physical verbs to developers: pick, place, and push. Because these grippers lack sensory feedback, the robot executes these actions blindly, unable to detect if an object slips or yields.

The NEO hand expands this vocabulary by offering 25 degrees of freedom, including 22 fully actuated joints in the fingers and palm, and three in the wrist. This mechanical design allows the AI model to access a much larger library of actions. Developers can program the robot to perform complex, delicate maneuvers like turning dials, tracing edges, and pinching small components. The hand behaves like a software library, providing the physical inputs and outputs necessary for the AI model to interact with diverse industrial and domestic environments.

This architecture shifts the challenge of robotic capability from hardware engineering to data collection. Instead of designing a custom end-effector for every new factory task, companies can use the same NEO hand for different applications. The hardware remains constant while the software model learns new skills through training data.

How does the 1X Tendon Drive enable read-write physical interaction?

The 1X Tendon Drive enables read-write physical interaction by operating at low gear ratios between 5:1 and 15:1, which minimizes internal friction and allows forces to flow bidirectionally. Traditional robotic hands are write-only devices because they rely on gear ratios of 100:1 or 200:1. These high gear ratios create significant internal friction. When the motor moves, the hand moves, but any external force pressing against the fingers is absorbed by the gearbox. The AI model remains blind to the resistance it encounters.

By contrast, the low gear ratios of the NEO hand offer force transparency. This means when an object pushes against a finger, the force travels backward through the tendon transmission directly to the motor. The system measures this resistance instantly. The AI model can "read" the physical environment by analyzing the force feedback from all 25 joints simultaneously.

This bidirectional flow transforms the hand into an active sensory instrument. When touching an object, the robot conducts a physical test instead of merely executing a position command. It measures material hardness, reads surface texture through fingertip friction, and detects weight. The hand is a perception stack that works alongside the robot's cameras to build a complete model of the surrounding space.

Force transparency versus traditional position control

In traditional position-controlled robots, the system only cares about reaching a coordinate in space, regardless of the resistance it meets. This rigid approach causes damage when a robot encounters unexpected obstacles or fragile objects. Force transparency allows the NEO hand to yield to external pressure. If a human bumps into the hand, or if the hand meets a solid surface, the joints give way naturally. The robot measures the contact force and adjusts its grip, preventing accidents and ensuring safe operation in shared workspaces.

Fast physical reflexes through proprioception

Proprioception is the robot’s internal awareness of its own joint positions and movements. Because every joint in the NEO hand is closed-loop, the AI model receives continuous updates on the exact pose of the fingers. This high-bandwidth data stream allows the system to execute fast reflexes. If an object begins to slide out of the robot's grasp, the system detects the change in shear force and joint position immediately. It can adjust its grip mid-action to catch the slipping object, matching human reaction times.

What are the business advantages of a software-defined robotic hand?

The primary business advantage of a software-defined robotic hand is the reduction of deployment costs and the elimination of hardware-specific integration cycles. In traditional automation, switching a production line to a new product requires physical retooling. Engineers must design, test, and install new grippers to handle different shapes or materials. This process takes weeks or months and halts production.

With the NEO platform, companies can adapt to new tasks through software updates alone. Because the hand matches human dexterity, it can handle the same tools, containers, and materials that humans use. A logistics company can retrain the robot to handle fragile electronics, heavy boxes, or irregular packages without changing a single physical component on the assembly line.

This capability simplifies the scaling of robotic fleets. Companies can train a single AI model on a specific task and deploy that software update to thousands of robots globally. The uniform, highly capable hardware ensures that the software behaves consistently across all deployed units.

Lowering the barrier to entry for enterprise AI integration

The NEO hand reduces the complexity of programming robotic systems. Historically, developers needed deep expertise in kinematics and control theory to program complex manipulations. The physical API simplifies this process by presenting the hand's capabilities as structured inputs and outputs. AI developers can focus on training high-level machine learning models using reinforcement learning and demonstration data, rather than writing low-level joint-control algorithms.

Increasing hardware durability and operational uptime

Industrial environments demand high reliability and low maintenance costs. Traditional robotic hands with complex, rigid gears suffer from wear and tear when subjected to repeated impacts. The tendon-driven design of the NEO hand is inherently robust. Because the joints are backdrivable and flexible, they absorb impacts instead of breaking. This elasticity reduces mechanical stress, decreases the frequency of hardware failures, and ensures that the robots maintain high operational uptime in demanding enterprise environments.

Key Takeaways

  • Transition to Software-Driven Automation: The 25-DOF NEO hand allows businesses to upgrade robotic capabilities through software model training rather than expensive physical retooling.
  • Deploy Safe Human-Robot Collaboration: Force transparency and backdrivable tendons allow the robot to sense and yield to external forces, ensuring safe operations alongside human workers.
  • Maximize Operational Uptime: The flexible, tendon-driven mechanics absorb physical impacts, reducing hardware wear and minimizing maintenance overhead in enterprise environments.