AUPA Risk32%
Limited authority-pressure evidence
YUDA Signal45%
Decision confidence reduced by actor uncertainty
Voice Stress Trend22%
No usable voice context available
Behavioral Deviation88%
Strong account behavior deviation
Score Methodology
How the Decision Integrity Score is derived. Weights below are illustrative and tuned with each institution during the pilot.
DIS = 100 − weighted( AUPA · YUDA · Stress Trend · Behavioral Deviation )
AUPA RiskHigh weight
External authority & pressure pattern
YUDA SignalHigh weight
Decision autonomy & answer consistency
Stress TrendMedium weight
Stress trajectory during confirmation
Behavioral DeviationMedium weight
Beneficiary / timing anomaly vs baseline
Decision Timeline
14:08:02
New device login
Device fingerprint not previously associated with customer.
14:08:18
Beneficiary added
New recipient created before transfer.
14:08:31
Navigation anomaly
Faster-than-normal flow across sensitive steps.
14:08:45
Escalation recommended
Step-up challenge and fraud escalation.
Explainability Panel
Engine Reason Codes
autonomy_degradedbehavioral_anomaly
Actor risk is high due to new device and session mismatch.
Decision integrity cannot be established because identity confidence is low.
Behavioral deviation is stronger than coercion pressure indicators.
Recommended action prioritizes account security controls.
Policy Note
Challenge the session, hold the transaction, and route to fraud operations.
Intervention Ladder
Action escalates with decision risk — friction stays proportional and hard blocks are the last resort.
70–100
Allow + Observe
Shadow-mode log only, no customer friction.
50–69
Soft Pause + Reconfirm
SMS / in-app reconfirmation on a safer channel.
Active tier for this scenario
30–49
Soft Pause + Care Call
Customer-care callback on a safe channel.
10–29
Temporary Hold
Hold transfer and notify a trusted contact.
0–9
Hard Block + Escalate
Block and route to fraud operations.
Pilot Report Preview
Example output a risk team would receive after a shadow-mode evaluation.
Synthetic demo only. No real customer data is used in this dashboard preview.
Open PDF-Ready One-PagerFinding
Coercion-like decision risk detected despite clean fraud status.
False-positive review
Hard block not recommended; start with soft pause and safer-channel reconfirmation.
Next step
Review 3-5 anonymized or synthetic journeys with fraud, compliance, and customer care teams.