Cleared without manual review.
Onboarding is the one moment where the institution has full control and the applicant's full attention. It is also the moment with the least information available. Bureau files are empty or stale. KYC confirms a document, not a person. Liveness confirms a face, not an intent. Shruhani enriches the onboarding moment with the signals no document carries, so the decision made at the gate is informed by the fullest possible picture of the applicant.
Indian financial institutions have digitised onboarding at scale. A loan application that previously required days now completes in minutes. An account opening that necessitated a branch visit now happens on a mobile device. That speed is no longer a differentiator; it is a competitive requirement.
The decisioning layer behind that speed has not kept pace. The same KYC, bureau and liveness checks that powered branch-era onboarding now sit behind digital-first flows — processing higher volumes, at greater speed, with fewer human touchpoints, and identical structural blind spots. A genuine applicant and a synthetic identity receive exactly the same verification checks. The difference between them lives in signals those checks were never designed to capture.
For each: the challenge, why current systems fail, how Shruhani addresses it — and what it measured in production.
Each check at the gate is necessary. None of them is sufficient, and none of them was designed to describe the person behind the application.
Shruhani scores that white space, in the same API call.
Phone, email, device, selfie, address and session behaviour are all present at onboarding — all capturable, all scoreable, and all resolved into a single applicant profile.
Cumulative outcomes from live onboarding deployments across banking, lending and credit cards.
100M+identities screened
Across onboarding, lending and AML decisions.
200K+fraud accounts intercepted
Before the first transaction.
₹800 Cr+in prevented losses
Estimated from customer reconciliation across deployments.
84%reduction in fraud approvals at onboarding
Leading fintech.
128%uplift in monthly disbursals
No increase in default rates. Lending client.
<200msp95 decisioning latency
Inclusive of network round-trip.
Indian financial institutions have digitised onboarding at scale. A loan application that previously required days now completes in minutes. An account opening that necessitated a branch visit now happens on a mobile device. That speed is no longer a differentiator; it is a competitive requirement.
The decisioning layer behind that speed has not kept pace. The same KYC, bureau and liveness checks that powered branch-era onboarding now sit behind digital-first flows — processing higher volumes, at greater speed, with fewer human touchpoints, and identical structural blind spots. A genuine applicant and a synthetic identity receive exactly the same verification checks. The difference between them lives in signals those checks were never designed to capture.
For each: the challenge, why current systems fail, how Shruhani addresses it — and what it measured in production.
A valid Aadhaar, a matching PAN and a face that passes liveness no longer confirm a genuine applicant. Complete identity kits are commercially available for ₹25,000. The documents are authentic. The individual presenting them may be a mule recruit, a synthetic identity, or a person whose credentials are being used without their knowledge.
KYC and liveness confirm the document and the face. They do not assess whether the individual has a genuine digital existence, a consistent device history, or a behavioural profile consistent with a legitimate applicant.
Shruhani evaluates the digital signature behind the application: footprint depth across 100+ platforms, device history and integrity, behavioural session patterns, and Face Vault matches against every historical application. Identity is what the documents establish. Intent is what they cannot.
78%
of fraudsters concentrated in the riskiest 5% of the applicant pool.
Personal loan provider
A first-time borrower applies for a personal loan. Bureau returns no score. The underwriting model receives no input and defaults to decline. The applicant may be entirely creditworthy — stable income, low expenditure, no existing obligations — but none of it is visible to a bureau-dependent model.
Bureau-derived scorecards need a minimum credit history to function. For the 300 million Indians with no bureau file, the model has no input to process. The system treats the absence of data as the presence of risk.
SMS Parser captures real-time financial behaviour: salary credit patterns, EMI regularity, spending volume, bounce frequency. Digital Footprint evaluates digital maturity and commerce depth. Location Intelligence adds affluence scoring at 100-metre resolution. Together they construct a credit-relevant profile where bureau has no coverage.
128%
uplift in monthly disbursals with no increase in default rates; ₹30 Cr unlocked in the 600–650 CIBIL band.
Lending client
In Tier 2 and Tier 3 markets, agents routinely assist applicants: filling forms, operating devices, capturing selfies on their behalf. The KYC documents belong to the applicant. The application was completed by someone else. In deliberate fraud, the applicant may not know an account or loan is being originated in their name.
Liveness confirms a live human is present. Face match confirms correspondence with the identity document. Neither confirms that the person operating the device is the applicant. Session behaviour, form-filling patterns and device handling are never evaluated.
Image Intelligence classifies the background scene — residential, lending kiosk, cyber cafe — and checks device metadata for consistency. Behavioural Biometrics detects copy-paste patterns, atypically rapid completion and coached navigation. Face Vault identifies the same face appearing under different names.
7.1%
of approved selfies identified as assisted captures; 1,400+ duplicate identity clusters across 19,986 KYC selfies.
Production review
Address verification in Indian banking has relied on physical field visits or postal confirmation — slow, costly and unreliable. A single pincode covers 60,000–70,000 residents, grouping neighbourhoods with entirely different characteristics. Verification at pincode level says almost nothing about the applicant.
Most address checks confirm that an address exists as a valid postal entry. They do not assess whether the applicant is reachable there, whether the address is consistent with their telecom registration, or whether the location’s economic profile matches the claims in the application.
Location Intelligence parses any Indian address format — incomplete, misspelled, mixed-language — geocodes it, and enriches it on a 100-metre hexagonal grid. Five composite scores are returned: affluence, economic activity, commercial activity, urbanisation and fraud risk. The address is not merely confirmed. It is contextualised.
100 m
satellite-based verification resolution, engineered for Indian address formats from a single address input.
Location Intelligence
Elevated false-positive rates are the hidden cost of onboarding fraud prevention. Every genuine applicant incorrectly flagged is a customer lost to a competitor. Every unnecessary step-up adds friction and abandonment. The institution pays twice: for the fraud it misses and for the legitimate applicants it rejects.
Single-signal checks operate as binary triggers. A rooted device is a flag. A recently activated number is a flag. A thin bureau file is a flag. Each one applies identical treatment regardless of what the remaining signals say. The result is a system that flags broadly and resolves imprecisely.
Shruhani evaluates five signal dimensions at once. A rooted device with a six-year WhatsApp history, a consistent session and a residential IP is not the same risk as a rooted device with a three-month number, no footprint and a proxy IP. Multi-signal scoring separates genuine risk from statistical noise.
84%
reduction in fraud approvals at onboarding with no increase in the false-positive rate.
Leading fintech
Each check at the gate is necessary. None of them is sufficient, and none of them was designed to describe the person behind the application.
Shruhani scores that white space, in the same API call.
Phone, email, device, selfie, address and session behaviour are all present at onboarding — all capturable, all scoreable, and all resolved into a single applicant profile.
Cumulative outcomes from live onboarding deployments across banking, lending and credit cards.
100M+identities screened
Across onboarding, lending and AML decisions.
200K+fraud accounts intercepted
Before the first transaction.
₹800 Cr+in prevented losses
Estimated from customer reconciliation across deployments.
84%reduction in fraud approvals at onboarding
Leading fintech.
128%uplift in monthly disbursals
No increase in default rates. Lending client.
<200msp95 decisioning latency
Inclusive of network round-trip.
We'll score it, surface the fraud patterns your current stack missed, and walk you through what we found. Your data, our scoring. No commitment, no integration, no decision required until you've seen the result.
