Claimed outcome (not L5-verified)
manual review rate: 62% → 21%
The number is the provider's claim. L4 verifies the technical work; outcome verification (with both attestations + an evidence-vault data source) is L5.
L4AI-analyzedmethodology vv1.0 · 2026-01
[Placeholder] Onboarding document classifier, Tier-2 bank
Production document classifier + structured extraction for onboarding KYC packets. Replaced a brittle rule pipeline.
By [Placeholder] Acme AI · sector: financial-services
Provider submission
The fields below are as submitted by the provider. The Trustgent AI analysis (further down) is a separate reading of these artefacts.
- Date
- 2026-01
- Sector
- financial-services
- Stack
- Anthropic, Postgres, AWS Textract for OCR
- Users
- 120 ops staff
- Data
- ~38k packets/mo
Trustgent AI analysis
The reading below is produced by methodology vv1.0. Each rubric criterion is scored with the AI's notes; a human reviewer signs off before publication. See how we analyze for the full rubric, prompt set, and model.
Technical depth
ReviewedThe provider's architecture matches the claimed capability and the named stack is production-grade. Confidence interval recorded with the analysis output.
Stack coherence
ReviewedThe provider's architecture matches the claimed capability and the named stack is production-grade. Confidence interval recorded with the analysis output.
Scale plausibility
ReviewedThe provider's architecture matches the claimed capability and the named stack is production-grade. Confidence interval recorded with the analysis output.
Complexity classification
ReviewedThe provider's architecture matches the claimed capability and the named stack is production-grade. Confidence interval recorded with the analysis output.
Claimed-outcome consistency
ReviewedThe provider's architecture matches the claimed capability and the named stack is production-grade. Confidence interval recorded with the analysis output.
Novelty signals
ReviewedThe provider's architecture matches the claimed capability and the named stack is production-grade. Confidence interval recorded with the analysis output.
What this page claims, and does not
This page documents the technical work the provider submitted and the Trustgent AI reading of it. It does not verify business outcome, and does not warrant current production status. For attested outcome see L5 records on the proof page.