Feather Robotics Builds the Body for Physical AI

Feather Robotics is building a purchasable body for physical AI. The East Palo Alto startup sells a modular bimanual robot with configurable compute, cameras, end effectors, and an open SDK so automation companies can focus on the application instead of reproducing an entire machine from scratch.

Founded in 2025 by CEO Hoa Mai and CPO Parsa Bakhtiari, Feather is entering the market at a moment when robotics models are becoming more capable but deployment remains stubbornly physical. A model can recognize an object or plan a movement; a working business still needs arms, power, safety, support, replacement parts, integration, and a machine that survives the customer's floor.

About Feather Robotics

Feather describes its mission as supporting the next generation of automation companies. Its target customers include physical-AI developers, manufacturing integrators, researchers, and businesses that need a configurable robot without committing years of engineering to a proprietary hardware stack.

The company occupies a specific place in the humanoid market. Vertically integrated robotics companies often control both the body and intelligence, while research systems can be accessible but may require substantial work before they are ready for commercial deployment. Feather is trying to offer a middle layer: configurable hardware, an open development environment, and field support at a price small teams can consider.

That strategy depends on more than making a robot cheaper. The platform has to be adaptable enough for many workflows without turning each deployment into a custom engineering project. Feather's modular design and software interfaces are intended to make that trade-off manageable.

How the Feather Platform Works

Feather's current robot weighs 105 kg and uses a holonomic base that moves in any direction without turning. A torso provides 600 mm of vertical travel, while two 1-meter arms each carry 7 degrees of freedom. Feather lists 23 total degrees of freedom, configurable Nvidia Jetson compute, optional wrist cameras, a touchscreen, audio hardware, an emergency stop, and support for custom end effectors.

The standard platform exposes an open Python and ROS2 SDK alongside teleoperation, inverse kinematics, collision detection, and simulation support. Dual hot-swappable batteries support the company's 10-hour operating claim, and the current limited-time starting price is $29,990.

These details matter because physical-AI teams face a stack of interdependent choices. A camera affects perception, compute affects latency and cost, an arm affects reach and payload, and batteries affect how the system fits a shift. Feather's product is designed to make those choices configurable without forcing a developer to become the manufacturer of every component.

Deployment Is the Product Test

Feather says its robots have worked in manufacturing, food service, and laboratory settings for more than a year. The company reports that it built its first prototype in 2 months, fulfilled its first order in 9 months, and has surpassed $1M in revenue. Customer names and deployment counts remain undisclosed, so those metrics should be understood as company-reported rather than independently audited.

TechCrunch's report describes robots performing restaurant cooking and laboratory cleaning tasks. Those use cases highlight the attraction and difficulty of the platform model. Different environments demand different tools, workflows, safety assumptions, and service responses, while the underlying body must remain dependable enough to justify reuse.

Feather's support model includes U.S.-based engineers, parts logistics, onboarding, a 1-year warranty, and fleet agreements. The company is treating service as part of the platform because an open SDK has limited commercial value if a broken machine leaves the application developer waiting for hardware support.

Leadership and Capital

Hoa Mai and Parsa Bakhtiari bring complementary backgrounds in embodied AI, humanoid systems, automotive production, and robotics. Their product choice reflects those disciplines: simplify the machine where possible, expose the configuration points developers need, and build support around real deployment.

Feather raised $7.6M in Pre-Seed SAFE financing in September 2026. Gradient Ventures led the round, with Builder Capital, Geometry, SEED Innovations, and Virgo VC participating. The financing gives the company room to expand production, engineering, support, and commercial execution while its early deployments test whether the platform can scale across customers.

What Feather's Hiring Says

Feather is currently recruiting a Forward Deployed Engineer, Full-stack Roboticist, Physical AI Researcher, and Associate Mechanical Design Engineer in the San Francisco Bay Area. That mix is a market signal rather than a generic hiring pitch. The company needs people who can move between research, mechanical design, customer environments, and the full software-hardware stack.

The open roles also show where the platform thesis gets difficult. A forward-deployed engineer brings customer problems back into the product. A roboticist has to make the complete system behave. A physical-AI researcher connects new models to the body, while mechanical engineering keeps configuration from eroding reliability. Feather is hiring at the seams where a platform either becomes reusable or becomes a collection of one-off projects.

Why Feather Matters Now

Physical AI is moving from model research toward commercial applications, but the path to a reliable machine remains expensive and fragmented. Feather is betting that a common hardware and support layer can increase the number of teams able to build businesses on top of robotics, much as accessible computing platforms expanded software markets.

The company still has to prove fleet reliability, repeatable production, and a developer ecosystem across more disclosed deployments. Its relevance comes from attacking a real constraint in the market: intelligence can improve quickly, while bodies, parts, field support, and customer trust continue to move at the speed of the physical world.