Will my pilot produce a defensible number?
Your pilot will produce a number. The question your committee will ask is not whether the number is good. It is who decided what counted as good, and when.
How pilots are run today
- The vendor picks the classical baseline it will be compared against.
- The compute budget on each side is never declared.
- The success criterion gets written after the results are in.
- A negative result quietly becomes a slide about lessons learned.
What a referee changes
We register and seal the success criterion before anything runs. We choose the classical champion — not your vendor — and we run it at the same declared budget. The result is published either way.
What you get
A sealed pre-registration, the classical champion’s full run at parity, a five-layer decision page, an optional ledger entry, and the frozen procedure@version — which means your next pilot starts from a protocol instead of from an argument.
Price
Pilot Referee — US$15,000–35,000 · 5–8 weeks.
Assumption stated plainly: typical corporate quantum pilots in our range run US$250,000 to US$1,000,000 all-in, and our invoice never exceeds 15 % of what it referees. If yours is smaller, tell us and we will scope down.
From the ledger, for you
We ran a quantum walk against five classical methods to find the binding pocket on KRAS G12C. It lost to every one of them, and then it lost to random guessing. We had sealed the criterion first, so we published it against ourselves.
The point of a referee is not to make your pilot succeed. It is to make its result mean something.