We make your pilot produce a defensible number.
Your pilot will end up in a vendor deck evaluating itself. We write the success criterion before it runs, race the champion classical solver on the same field, and seal the result — a number nobody can rewrite, including us.
How a referee works
We write the success criterion BEFORE anything runs, and seal it. After that it cannot be adjusted by looking at the result — not by us either.
The strongest classical solver runs the same instance on the same compute and time budget. Without that, the comparison does not measure what it claims to.
The result is hash-anchored to Bitcoin via OpenTimestamps and mirrored in three places. The record cannot be rewritten.
A one-page decision page plus the five-layer report: verdict, margin, robustness, what would flip it, and what it means in your units.
Price
$15,000–35,000 per pilot, and never more than 15% of what the pilot costs. If our invoice weighs on the decision, the referee stopped being a referee.
Who does NOT pay us
The vendor under evaluation does not pay for the verdict, directly or indirectly. They can present, they can co-sell, they can be in the room — but the corporate contracts and pays. That is the Two-Layer Rule, and it is the only reason the number is worth anything.
The same physics that ran in the sealed experiment.
KRAS G12C, structure 4OBE, with the contact network read from the sealed topology. The walk runs in your browser on the same code as the experiment; what the verdict rests on is the seal, not this. Source residues are teal and predicted sites red — different colours on purpose, because the experiment measured that they do not coincide.
Five industries, five sealed experiments
Pharma, finance, energy, banking and aerospace. In all five the classical champion ran on the same field. In all five we published the result, and in none of them did the quantum side win — yet.