Mech-1 · shipping
The co-processor
The safety core on a certified ASIL-D board, wired onto open platforms. Available now for design partners.
The chip
Mech-1, the safety core.
Mech-1 is a single chip that gives a humanoid robot its boundaries. It sits beside the main processor, watches every movement the AI plans, and clears it against your rules. It runs on certified safety silicon, and you can design it in today.
Mech-1 on a reference board. Certified silicon, ready for design partners.
How it works
Between the model and the motors.
The AI plans a movement. Before any power reaches a motor, the plan crosses Mech-1. The chip checks it against your rules and clears it or holds it. The decision happens on the chip, in real time.
- [1]
Its own power path
Mech-1 controls power to the motors directly. The safety decision stays independent from the AI model.
- [2]
Fixed and checkable
The safety logic is set and formally verified. It behaves the same way every time.
- [3]
Signs its work
Every decision is written to a sealed log on the chip's own identity.
The line
Mech-1, and what comes next.
We ship the safety core on proven silicon today, and each generation moves closer to a die of our own.
Mech-2 · next
The module
A certified module robot makers design in, with the seal on the box. The mark buyers ask for by name.
Mech-3 · roadmap
The die
Our own safety die, built fabless. The heavy compute stays a commodity others supply.
Mech-1
What is on the board.
Safety rating
ASIL-D target, on automotive-grade safety silicon
Decision path
Mech-1 controls power to the motors directly, in hardware
Decision time
Under 1 millisecond, on the chip
AI model
Yours to choose, and yours to swap any time
Identity
A hardware-rooted Sequence that signs the robot's logs
Platforms
Open-SDK humanoids, Unitree G1 supported first
Network
Runs fully offline
The seal
Each sealed unit carries the Mech mark, simple for a buyer to verify
Get a board
Design Mech-1 in.
Tell us about your platform. We will scope the integration and send a board.