Clinical AI translation
you can take to court.

We don't ship confidence scores because confidence scores are self-reported. We ship verifiers. A verifier is a deterministic test function that evaluates one outcome property of a translated utterance — was the dosage preserved? does the register match what the clinician requested? did the output mention a name from the exclusion list? Each verifier returns pass or fail with a structured reason, and carries an explicit weight from +5 (catastrophic) to +1 (stylistic).

The production suite contains approximately 60 verifiers organized by clinical specialty and register. The full methodology — principles, taxonomy, version control, escalation rules — is documented in Section 4 of the whitepaper. Below is one representative verifier from Appendix A, executable as-is in a pytest harness.

Adapted from outcome-based rubric work in OpenClaw Atlas and HiL-Bench · read the full methodology →

# From Orquor Clinical verifier suite v1.2
# Appendix A · ACTO whitepaper v0.1

def verify_dosage_preserved(
    source: str,
    translated: str,
) -> VerifierResult:
    """Numeric dosage in source must appear
    identically in translated output.
    Weight: +5 (catastrophic if violated)."""

    source_dosages     = extract_numeric_dosages(source)
    translated_dosages = extract_numeric_dosages(translated)

    missing = source_dosages - translated_dosages
    if missing:
        return VerifierResult(
            passed=False,
            weight=5,
            reason=f"Dosage {missing} missing",
        )
    return VerifierResult(passed=True, weight=5)