Launch report
Introducing the Public AI Documentation Index
ModelCompliance.ai reviews public documentation from B2B AI SaaS companies and turns scattered buyer-facing evidence into structured, source-linked transparency profiles. The index focuses on what a public buyer, researcher, or partner can find without private access.
V1 tracks public evidence of AI usage disclosure, model/provider disclosure, customer-data training statements, subprocessors or AI vendor disclosure, human oversight statements, AI safety/testing or evals statements, responsible AI/governance policies, trust centre presence, and references to frameworks such as EU AI Act themes, NIST AI RMF themes, and ISO/IEC 42001 themes.
The index does not determine legal compliance, does not certify companies, and does not assess private controls. Profiles may be incomplete or outdated because public documentation changes and some evidence is not public. Each profile therefore emphasises source links, confidence, last-reviewed dates, and plain evidence statuses such as found, partial, not found, unclear, and not applicable.
Corrections and profile claims are free. Public profiles are updated based on supporting public evidence, not payment. The evidence-pack request workflow is a separate documentation/readiness support path for teams that want help organizing buyer-facing AI evidence before review and publication decisions.