For manufacturers

Today, humanoids deploy where trust was built one partner at a time: a shareholder in the cap table, years of joint pilots, an industrial partner with its own stake in the outcome. Schaeffler is the clearest case: minority investor in Agility, deployment and actuator deals with Hexagon and Humanoid, four-digit fleets planned in its own plants. It deploys the humanoids and supplies the joints they run on: both sides of the table.

Your next hundred customers hold no shares, ran no pilot with you, and will not take your word for the state of your machines.

You know better than anyone what a long pilot costs in engineers on site and revenue pushed back. The blocker at the end is rarely the humanoid. It is the proof around it: the customer’s insurer, its works council, its safety officer, and the next customer who was not in the room. None of them can accept the seller grading its own machines.

Type testing, homologation, update validation on reference units: yours. We never touch it, the same way the vehicle inspection does not redo the crash test.

Our object is the unit in service: does humanoid no. 047, with 4,000 hours in that plant, still hold the specifications you declared? Industry already treats every machine that runs near people this way. An elevator is certified at build; in Germany an approved independent body still inspects it every year and re-triggers the safety gear, and after an accident, that body’s safety assessment is what the law asks for. Nobody reads that as distrust of the elevator maker. It is what lets elevators be everywhere.

On cars the two have coexisted for thirty years, and the modern test lane plugs into the OBD and cross-checks it. We work the same way: your fleet data is an input we anchor to an external calibrated reference, not a rival we ignore.

Direct joint commands, AI correction off, read access to motor currents and temperatures. With it, our station measures your humanoids in open loop: the naked mechanical truth, before the controller compensates it.

A real layer of trust with your customers. Humanoids that clear the insurer and the works council sell faster, and stay deployed. And something your telemetry structurally cannot produce: independent, calibrated wear data from worn units in real sites, measured against an external reference that survives your updates. Your engineers get ground truth from the field. Your customers get proof.

We believe every deployed humanoid will end up needing independent in-service proof, and that operators will write it into purchase orders: their insurer, their works council and their own legal duties all push in that direction.

That proof cannot exist without two objects: a test protocol per humanoid model, and a test interface in the humanoid’s software.

The maker who opens first gets three compounding advantages. The reference protocol is shaped on your humanoid: your machines are verifiable from day one, competitors retrofit later. Your fleet starts its verified history first, and a measurement history is the one asset a latecomer cannot replay. And when insurers and buyers start referencing a protocol in policies and purchase orders, they reference the one that exists: the one written on your machines.

That is what defining the standard means here. Not a committee seat: the physical protocol the market copies. The OBD port is the precedent: one test interface, defined once, that the whole inspection world now plugs into.

Once one maker opens, refusing starts to look like hiding.