An exchange-rate oracle
for AI agent labor.

Empirical reference rates for AI agent classes. Patent-pending methodology. Currently in private testing with eight frontier models across five task domains.

What it looks like

Live exchange rates and quality scores updated daily across Claude, GPT, Gemini, and Grok families. Full preview behind authenticated access.

How it works

Cross-platform blind grading
Each platform scores its peers without knowing whose output it's grading. Self-grading bias is removed by construction.
Multiplicative capability coupling
Efficiency acts as a coupling constant on the base value, not an additive factor. Doubling efficiency more than doubles value.
Adaptive sampling
Per-platform, per-domain sampling rates self-adjust based on confidence interval width. Stable performers get sampled less; volatile ones more.
Real-time exchange rates
Pair rates derive from rolling ALU values, FOREX-style. Each platform always has a current rate from its rolling EWMA.

Intellectual property

Two provisional patents filed at the United States Patent and Trademark Office — April 16, 2026.

Co-inventors: James Gill, David Schwartz.
Joint IP Ownership Agreement executed.
Non-provisional conversion pending Q1 2027.

Request access

Authenticated visitors can view a fuller methodology preview with redacted sample data. Investors and IP attorneys: please include your firm in the message body.

Send Access Request

Or write directly to [email protected].