Work with us
Want to measure — or verify — quantum hardware?
Design-partner pilots open now; certification follows the data.
ethan@admissibletech.com
Every quantum vendor grades its own homework, and every buyer takes the results on faith. Our engine measures hardware the way an operator measures a fleet — and answers the question the field keeps fighting about: is this device performing to spec, and is this workload genuinely hard for a classical machine?
We replace the one number with a diagnosis. For every classical explanation of what the device did, we weigh the two prices the diagram plots — cost, the classical resource it must spend (including the hidden bookkeeping a classical model has to invent), and distortion, the error that explanation has to accept. Where the cheapest classical explanation lands against the budget sorts the device into a four-way verdict: healthy and in spec, near a limit, drifting, or genuinely quantum. That last verdict has a hard backstop — once a device violates a Bell test, no classical model can match it, and the result can't be argued down.
The output of this assessment is the product: a four-way health read with a live drift monitor, and a signed certificate that a workload was hard for a classical machine. It's vendor-agnostic and runs on the data the hardware already emits.
Provisional filed (App. 64/070,738).
For any quantum interface we compute two numbers: the cost of reproducing it with a classical (Boolean) model, and the distortion that model introduces. Plot them, and the hardware lands in one of four regimes. Inside the admissible rectangle — low cost, low distortion — a classical machine can keep up. Outside it, the device is doing something genuinely quantum. The verdict comes with a signed margin, so it's robust to calibration jitter, not a coin-flip at the boundary.
The same engine runs as a live monitor. It watches a device's operating telemetry with anytime-valid statistics — alerts that stay sound no matter when you look — flags drift before a run is wasted, and decomposes an excursion into why it happened. Run it across many machines and you accumulate the one asset no one else has: comparative, cross-fleet measurement. That corpus is what turns observability into the neutral certification authority — the "UL / Moody's for quantum compute."
The cost-distortion frontier, the four-regime verdict, the classical-embedding test, and the anytime-valid wrapper are grounded in the company's published technical papers. The engine already exists inside our verification codebase, and an MVP runs daily against publicly accessible hardware — productizing it is consolidation, not invention from zero.