Business correspondent device
Legitimate assisted onboarding creates expected device reuse, so thresholds must understand the channel.
Keep assisted onboarding open without flying blind.
Add personhood evidence to branch, business-correspondent and digital account opening while preserving the institution's existing KYC controls.
Channel context changes the evidence.
A shared device at a branch means something different from a shared device across remote applications at midnight.
Repeated beneficiary
A common payout or transfer destination across newly opened accounts can reveal operator control.
Rural thin file
A first-time borrower may have little bureau evidence and still show a coherent, independent identity.
Post-opening mule use
Transaction monitoring finds movement after opening; application scoring adds evidence before the account exists.
32.08 lakh
Layer-1 mule accounts shared through the I4C Suspect Registry
Ministry of Home Affairs, Lok Sabha reply, 30 June 2026
Financial inclusion fails when a thin file is treated as evidence of a false person.
Calibrate by channel before taking action.
The same signal can support, weaken or remain neutral depending on whether onboarding was assisted, branch-led or fully remote.
Personhood Score
Origination
Use channel-calibrated thresholds and reason codes at account opening.
V-CIP Integrity
Remote channel
Check remote video sessions for generated media and repeated identity vectors.
Cluster Intelligence
Network layer
Separate expected branch device reuse from suspicious cross-branch application infrastructure.
Workbench
Operations
Route linked cases to central fraud teams with a reproducible evidence record.
Backtest by branch and channel.
The analysis segments historical applications by branch, correspondent, remote channel and product. That prevents a busy assisted-onboarding device from being judged by the same baseline as a consumer handset.
Application scoring complements existing monitoring.
The RBI advisory of 30 June 2025 directs covered banks to integrate DoT's Financial Fraud Risk Indicator. Praman Labs consumes permitted signals at origination and does not replace FFRI or MuleHunter.AI.
Source: Department of Telecommunications via PIB, 30 June 2025
Run the Banks and small finance banks backtest.
The first output is a ranked historical file and a precision curve at your available review capacity. No production integration is required.
Book a backtest