Zeroth, a Suzhou-based AI robotics startup, has launched a compact humanoid called Bridge alongside an open developer ecosystem called OpenBridge. RoboticsTomorrow reported that the company introduced both on September 2 and opened global booking for the Bridge Geek Edition.
The launch is designed to pair a physical robot with tools for outside developers. RoboticsTomorrow described Bridge as about 88 centimeters tall and 13 kilograms, with 30-to-40-centimeter arms and two-finger grippers. It said the robot offers access to SDKs, open motion-control APIs, and motion-capture and VR integration. OpenBridge is intended to give creators a common toolset, pooled training material, and a publicly accessible Skill Hub.
Zeroth calls Bridge the world’s first AI-native embodied robot. That is a company claim, not an independent industry classification. The more verifiable point is that the company is launching a small humanoid while simultaneously inviting developers to create and share software capabilities around it. In a crowded Chinese robotics market, the effort to build an ecosystem may be as important as the hardware itself.
Bridge Pairs a Compact Body With Developer Access
RoboticsTomorrow describes Bridge as a compact humanoid designed for real-world development and creator use. The company’s announcement says it has stable bipedal walking, obstacle avoidance, a library of movements, and interfaces for custom algorithm deployment and scenario tuning. Those are product descriptions from the launch material. They do not substitute for independent tests of reliability, battery life, safety, or performance in commercial settings.
The dimensions show why Zeroth is taking a different route from full-size industrial humanoids. An 88-centimeter, 13-kilogram machine can be easier to move, deploy, and test in smaller spaces than a heavier human-scale robot. The report says Bridge has arms between 30 and 40 centimeters long and two-finger grippers, details that point toward simple manipulation and research use rather than a claim that it can perform every household or factory task.
The development interfaces are central to the launch. SDK access and open motion-control APIs give an outside developer a way to change behavior or connect their own software. Motion-capture and VR integration can provide another route for demonstrating or collecting movement data. The source does not disclose the full software license, hardware price by region, or what level of access each buyer receives, so those details should not be assumed.
China’s robotics sector has increasingly recognized that hardware is only part of the proposition. EastFrontier’s report on China’s robot games platform as embodied-AI infrastructure examined the need for data, training environments, and software layers around physical machines. Bridge and OpenBridge fit that wider shift by presenting a robot body together with a set of development pathways.
OpenBridge Attempts to Make Robot Skills Shareable
OpenBridge is Zeroth’s response to a familiar robotics problem: software created for one machine or team is often difficult to reproduce elsewhere. RoboticsTomorrow says the platform offers a single engineering environment, pooled material for training, and an open Skill Hub where contributors can exchange robot functions.
The model is explicitly open-source in the company’s framing. The report says Bridge natively supports OpenBridge and that the platform is meant to let developers co-build and share capabilities. Openness can attract experimentation because a creator can inspect interfaces, develop a new skill, and make it available to others. It does not by itself ensure compatibility, safety, quality control, or a large community of active contributors.
Zeroth has also launched a Bridge Builder Program. The first phase will shortlist 500 developers, according to RoboticsTomorrow. Two hundred will receive complimentary access to Bridge and participation in the OpenBridge co-creation effort, while 300 will get priority purchase access, SDK access, and eligibility for later openings. The count is a precise description of the announced program, not a measure of how many developers have already built or deployed applications.
The incentive design makes the company’s strategy clear. A robot manufacturer can sell units, but an ecosystem becomes more valuable if people build applications that a manufacturer did not anticipate. OpenBridge is meant to make that process more visible and reusable. Whether it attracts the right developers will depend on documentation, hardware availability, cost, and the quality of the platform’s tools.
The approach sits alongside a broader push for standards and shared systems in China’s humanoid industry. The country has already set a 2028 standardization test for humanoid robots, while individual companies are experimenting with their own hardware, models, and development stacks. Zeroth’s contribution is not a national standard. It is a proprietary hardware product paired with an open ecosystem that aims to encourage outside creation.
The Launch Tests an Ecosystem Strategy in Robotics
Zeroth was founded in 2024 and is headquartered in Suzhou, according to RoboticsTomorrow. The announcement describes work across core components, motion control, and embodied AI models. It says the company is pursuing companion robots, interaction systems for several settings, and systems for complex environments. These are company descriptions of its focus, not evidence of proven large-scale deployment.
The company’s choice to debut Bridge with OpenBridge gives it a way to differentiate from robotics firms that mainly promote a body, a demonstration, or a funding round. Zeroth enters a market where investors are also looking beyond the body itself toward the software and data layers, a shift examined by EastFrontier in its report on embodied-AI investors moving from bodies to brains. The September launch provides a concrete step: an announced product, a booking route, and a developer program.
The limitations are equally clear. The launch announcement does not provide independent benchmarks against rival humanoids, customer contracts, production volumes, or reliability data. It also does not show whether the Skill Hub has active third-party contributions. Those will be more meaningful tests of the ecosystem than the number of people who apply for the Builder Program.
For now, Bridge is a compact humanoid with a defined hardware form and a stated aim to let developers customize it. OpenBridge is the bet that robot skills can become reusable community assets rather than isolated features. If Zeroth can make that work, it may gain a network effect that hardware specifications alone cannot provide. If it cannot, the company will still need to compete on the physical performance and practical usefulness of Bridge itself.