Free to measure. Pay to certify.

Free to measure. Pay to certify.

One calibrated ruler measures every performer: human, AI agent, robot. One interval-scale metric. Stated uncertainty.

Score anything for free. Every score comes with a standard error on an interval scale, the same property NIST AI 800-3 names directly: "the intervals between LLM capabilities on the latent scale have meaning independent of the difficulties of the specific tasks selected into the benchmark. Raw benchmark scores do not have this property."

When you need tighter uncertainty, you move down the ladder. Each tier below is PROPOSED pricing under active test; it is not final and not yet contractual.

PROPOSED tiers:

• SE 0.30 screening and coaching: $150 to $400 per analysis.
• SE 0.20 default tier: $1,500 per analysis.
• SE 0.14 defensible tier: $3,000 to $5,000 per analysis, built for decisions that must survive audit.
• SE 0.10 certification tier: $250,000 to $750,000 per year.

The ceiling is SE 0.07, demonstrated on the AIM ruler and available in limited lighthouse slots. PROPOSED: allocation is capped while we validate throughput at that precision. If your decision requires publishable measurement-grade certainty, request a slot now or wait for capacity.

Every analysis runs on a unidimensional Rasch hallucination construct, so scores mean the same thing across performers and over time. Our Coaching Supervision stack is Rasch-analyzed end to end. Chapter 18 of the Cambridge Handbook covers our unobtrusive measurement approach: no prompts engineered to catch errors, no behavior changed by being watched.

Start with free scoring. See your number and its uncertainty before you spend anything. Move down the ladder only when the decision in front of you demands it.

Button: Start Free Scoring

Supporting line: No card required. Your first interval-scale score and its standard error are free. When you are ready to certify rather than screen, request a Defensible Tier quote.