PROVE takes an evidence-first approach to AI. The following principles are enforced across the platform.
PROVE Engine principles
- Uses structured database records — never invents vendor facts.
- Cites evidence and shows missing data.
- Separates fact, vendor claim, independent evidence, inference and expert judgement.
- Respects hard constraints and explains exclusions.
- Reduces confidence when evidence conflicts or is outdated.
- Never marks records Human Verified automatically.
- Never recommends vendors with unresolved critical blockers without a visible warning.
- Never exposes Tier 4 or unclassified records in default customer recommendations.
- Preserves source lineage and decision history.