One normalized risk score per document — backed by per-field forensic evidence.
One POST returns AI-generation indicators, tampering regions, metadata drift, math reconciliation, and a verdict — designed for risk engines, not onboarding funnels.
Inspects EXIF, software signatures, edit history, and structural fingerprints.
Originals are processed in encrypted memory and removed after analysis. Reports stay redacted by default.
Engineering teams need a single normalized number to feed into decisioning — but they also need the per-field evidence behind it for review and audit.
TrueDoc returns a trust score for every verified document, alongside the structured evidence map that drove it.
Hard yes/no verdicts that hide nuance
Black-box scores with no underlying evidence
Scores that don't generalise across document types
Hidden recalibrations that break downstream thresholds
Send the document; the response includes a trust score plus findings.
Normalized number combining forensic, structural, and AI-generation signals.
Per-field findings explain the score — useful for review and audit.
Auto-accept above your threshold, route mid-range to review, escalate the lowest scores.
Every /verify response includes a trust score, an ordered list of findings (with confidences and field pointers), an AI-generation indicator, and a stable verdict ID.
The verdict ID links to the immutable record in the enterprise portal — useful for review and audit.
Auto-accept high scores in low-risk workflows to reduce manual review volume.
Route mid-range scores to a reviewer queue with the evidence pre-attached.
Trigger step-up checks (additional document, biometric, manual review) only on the lowest scores.
No score is a 100% verdict. Treat it as a strong signal, not a single source of truth.
Re-encoded, low-resolution, or heavily compressed files reduce signal strength and may lower confidence.
No credit card. Redacted report in under a minute.