Score before
you trust.

The AI-powered API for identity, income, and fraud intelligence. iScoreIt transforms complex data into predictive risk scores. Evaluate every applicant with AI-driven precision and make confident, defendable decisions in under a second.

Book a demo Explore the APIs

drag the globe

Built for banks, NBFCs, fintechs and marketplaces

Atlas MarketsPeridot CapitalOnyx InsureMeridian GamingTrident TelecomNorthline BankSahaj Finserv

0+

Verification APIs

<0ms

Median response

0%

Platform uptime

0 key

UAT and live, one console

Who it's for

The same platform, tuned to your risk.

The signals overlap · the policy does not

Underwrite thin-file borrowers without asking for a single document.

Surrogate and bureau signals give you an income estimate, an ability-to-pay view and a fraud read before the applicant ever uploads a payslip.

  • Document-less estimated income for salaried and self-employed profiles
  • Bank statement analysis with salary, obligation and bounce detection
  • First-party fraud checks on device, mobile vintage and email footprint
  • Policy changes shipped from the console, not a release cycle
See lending APIs
Decision trace · APP-88214retail-pl-v14
01

Application received

PAN, mobile, consent captured

0 ms
02

Identity & bank verified

Name match + reverse penny drop

190 ms
03

Income estimated

Surrogate model, no upload

340 ms
04

Decision returned

Approve, limit ₹1.87L

412 ms
Elapsed on this request
0.0s
Median 4 s is illustrative, pending measured data.
01

One request enters — through the API door — Console and VerifyLink land in the same place.

02

Your policy loads — v3.2, versioned, owned by risk.

03

Checks fire in parallel, each against the issuing source.

04

One source stalls. Failover reroutes it — the funnel never notices.

05

Signals become one score, weighted by your policy.

06

Decision out — approve, with reason codes, consent artefact and audit trail attached.

01 Delivery surfaces
Same checks, same trail
REST API
Console
VerifyLink
02 Decision engine
Your policy, versioned
policy v3.2 loaded
03 Verification
40+ checks, in parallel
AadhaarDigiLocker
PANNSDL
Face matchEngine
BureauCIBIL
Mule screenRegistry
Bank a/cRails Avia B
04 Source & resilience
Issuing-source first, failover behind it
Source timeout — rerouting
DigiLocker
NSDL
CIBIL
Registry
Bank rails A
Bank rails B
REQ-88214
KYC + risk score · one key
ResponsePOLICY v3.2
0/100
Risk score
Approve
kyc.source_matchincome.est_okdevice.clean
Consent artefact
Reason trail
Audit trail
One request in. One decision out. 4.0 s.
Keep scrolling
Built to survive an audit

Degrade gracefully, never silently.

A verification call that fails closed can stop your business; one that fails open can cost you far more. iScoreIt does neither — it tells your policy exactly which signal is missing and lets you decide.

Per-signal status

Every check returns its own status, not an all-or-nothing response — a slow source never sinks the whole call.

Automatic failover

Where more than one upstream source exists, traffic reroutes on its own and the journey never notices.

Replay & audit

Idempotency keys, versioned policies and a stored consent artefact make any decision reproducible months later.

ISO 27001SOC 2 Type IIDPDP-alignedData residency · IndiaAES-256 at rest · TLS 1.3
Products

Three products. One decision surface.

// 01 · Decide

Pulse

The decision engine your risk team can edit themselves — policies versioned on save, shipped without a release cycle.

Talk to us about Pulse
// 02 · Onboard

Verify

Onboarding journeys that stop asking once they are sure — risk-based step-up, hosted or headless.

Talk to us about Verify
// 03 · Protect

MuleNet

The money-mule map, built from what fraud actually publishes — payout accounts screened before the money moves.

Talk to us about MuleNet
// 04 · Speak

Dialect

Voice agents that verify while they talk — twelve Indian languages, identity confirmed mid-call, every conversation scored and filed.

See how Dialect works
One identity graph · One policy layer · One audit trail
iScoreIt Labs

What we learn from the fraud we see.

Our team publishes what the data shows — how mule chains are actually structured, which signals hold up under pressure, and where verification quietly fails.

No vendor gloss

Mule networks

Anatomy of a mule chain: how deposit money moves through hundreds of accounts in a single afternoon

Six-week field studyRead the study

Thin-file credit

Estimating income without documents: which surrogate signals actually predict repayment

Out-of-time testedRead the study

Device intelligence

The shared-device problem: when account linkage is fraud and when it is a family

Threshold guideRead the study

Customer · Head of Credit Risk

“We were running six vendors and reconciling them by hand. iScoreIt collapsed that into one call and one policy — and the review queue dropped by more than half in the first month.”
CR Head of Credit RiskMid-size NBFC

AI RISK INTELLIGENCE · KYC & FRAUD APIS FOR INDIA

IDENTIFY

Score before you trust. One API layer for identity,

income and fraud intelligence — 40+ signals · one request.

POST /v1/DECISION · 200 OK · 412 MS

SCORE

Every applicant, merchant and payout account becomes

a decision your team can defend — iScore 782 · low risk · approve — in under a second.

MULENET · CONTINUOUS FRAUD DISCOVERY

SECURE

MuleNet maps money-mule networks from the deposit pages fraud publishes,

and screens your accounts before the money moves.

Book a demo Explore the APIs

SANDBOX KEYS THE SAME DAY · DATA RESIDENCY IN INDIA · 99.9% UPTIME

Products

Pulse decides. Verify onboards. MuleNet protects. Dialect calls.

Three products on one identity graph, one policy engine and one audit trail — so a decision made anywhere is explainable everywhere.

01 · Decide

Pulse decision engine

The decision engine your risk team can edit themselves.

  • Visual policy builderno deploy
  • Shadow & champion–challengersafe rollout
  • Decision replayaudit
  • Cost-aware waterfallsspend control
Talk to us about Pulse
02 · Onboard

Verify onboarding suite

Onboarding journeys that stop asking once they are sure.

  • Hosted or headlessyour call
  • Risk-based step-upless drop-off
  • UAT and live in one consoleone key set
  • Re-KYC & periodic reviewstays current
Talk to us about Verify
03 · Protect

MuleNet fraud intelligence

The money-mule map, built from what fraud actually publishes.

  • Continuous discoveryalways on
  • Screenshot-backed evidencedefensible
  • Linkage scoringbeyond exact match
  • Payout-time screeningbefore money moves
Talk to us about MuleNet
04 · Speak

Dialect voice agents

Voice agents that verify while they talk.

  • Twelve Indian languages, code-mixedswitches mid-call
  • Identity confirmed inside the callnot assumed
  • Conduct guardrails and calling windowenforced
  • Every call scored, not sampledQA on 100%
See how Dialect works
Book a demo See every API
Solutions

Built for the teams who carry the risk.

Lenders, fintechs and compliance teams run the same platform with different policies — same signals, same audit trail.

For lenders

Underwrite thin-file borrowers without asking for a single document.

Surrogate and bureau signals give you an income estimate, an ability-to-pay view and a fraud read before the applicant ever uploads a payslip.

  • Document-less estimated income for salaried and self-employed profiles
  • Bank statement analysis with salary, obligation and bounce detection
  • First-party fraud checks on device, mobile vintage and email footprint
  • Policy changes shipped from the console, not a release cycle
See lending APIs
Decision trace · APP-88214retail-pl-v14
01

Application received

PAN, mobile, consent captured

0 ms
02

Identity & bank verified

Name match + reverse penny drop

190 ms
03

Income estimated

Surrogate model, no upload

340 ms
04

Decision returned

Approve, limit ₹1.87L

412 ms
For fintechs & platforms

Onboard merchants in minutes, and keep mules out of your payout rails.

KYB, beneficial ownership and payout-account screening run in the same call, so a seller goes live the same day without opening a hole in settlement.

  • GSTIN, MCA, Udyam and shop licence checks with director linkage
  • UPI ID and bank account verification before the first payout
  • MuleNet screening against collection accounts seen in the wild
  • Ongoing monitoring — accounts re-scored, not just checked once
See MuleNet
Onboarding trace · GSTIN + PANMuleNet on
01

Merchant signs up

GSTIN + PAN submitted

clear
02

Entity graph built

3 directors, 1 linked entity

review
03

Payout account screened

Not seen on collection pages

clear
04

Live with monitoring on

Re-scored every 24 hours

active
For risk & compliance

AML screening and an audit trail your regulator can actually read.

Every check, every model version and every override is stored against the case. Replay any decision from six months ago exactly as it was made.

  • Sanctions, PEP and adverse media screening with tunable match thresholds
  • Full decision replay — inputs, policy version, outcome
  • Maker–checker on policy changes, with diff history
  • Data residency in India, consent artefacts stored with every call
Read the trust overview
Case file · DEC-88214Policy v14 · model v3.2
01

Case opened

Ref DEC-88214 · 14 Mar

logged
02

Screening run

Sanctions, PEP, adverse media

1 hit
03

Analyst disposition

False positive, note attached

closed
04

Replay available

Inputs, policy and model frozen

frozen
Talk to us about your cohort
APIs

Forty-plus checks. One key. One catalogue.

Send an applicant, name a policy, get a scored decision back with every underlying signal attached — sandbox and live behave identically.

Identity

PAN, voter ID, driving licence and passport — plus Aadhaar and every other document pulled straight from DigiLocker, on your own requester registration. Face match, liveness and OCR handle whatever still arrives as an image.

DigiLockerPANAadhaarFace matchLivenessOCR

Business & KYB

GSTIN, CIN and MCA filings, Udyam, shop & establishment licences, director networks and beneficial ownership — resolved into one entity graph.

GSTINMCAUdyamUBO

Banking & payments

Penny drop, reverse penny drop over UPI, UPI ID validation, IFSC resolution and bank statement analysis with salary and obligation tagging.

Penny dropReverse PDUPIBSA

Signal intelligence

Device fingerprinting, email existence and digital footprint, mobile vintage and porting history, IP and geo risk — the quiet signals fraud can't fake cheaply.

DeviceEmailMobile ageIP risk

Income intelligence

A monthly income estimate built from surrogate and bureau signals. No payslip, no statement upload, no drop-off in the middle of your journey.

Document-lessSurrogateBureau

AML & screening

Sanctions, PEP and adverse media screening with tunable thresholds, ongoing rescreening, and dispositions stored against the case for audit.

SanctionsPEPAdverse media

The module that matters

Fraud & mule intelligence

MuleNet continuously discovers illegal betting and gambling deposit pages, captures the collection UPI IDs, account numbers and QR codes shown on them, and screens your accounts against that evidence repository.

MuleNetUPI watchlistLinkage

The full catalogue

POST/v1/decision
POST/v1/mulenet/screen
POST/v1/income/estimate

26 endpoints, grouped by the question they answer.

Get sandbox keys

Built for developers

  • Idempotent requests and signed webhooks
  • Sandbox that returns the same shape as live, with seeded test identities
  • SDKs for Node, Python, Java and PHP
  • Every response carries the policy and model version that produced it
Get sandbox keys Open the live console
One endpoint

One endpoint. Every signal.

Send an applicant, name a policy, get a scored decision back with every underlying signal attached. No orchestration code, no vendor-by-vendor error handling, no reconciliation job.

  • Idempotent requests and signed webhooks
  • Sandbox that returns the same shape as live, with seeded test identities
  • SDKs for Node, Python, Java and PHP
  • Every response carries the policy and model version that produced it
Insights · iScoreIt Labs

What the data shows.

Our team publishes what the data shows — how mule chains are actually structured, which signals move approval rates, what income estimation gets right and wrong.

Mule networks

Anatomy of a mule chain: how deposit money moves through hundreds of accounts in a single afternoon

Six-week field studyRead the study

Thin-file credit

Estimating income without documents: which surrogate signals actually predict repayment

Out-of-time testedRead the study

Device intelligence

The shared-device problem: when account linkage is fraud and when it is a family

Threshold guideRead the study

Customer · Head of Credit Risk

“We were running six vendors and reconciling them by hand. iScoreIt collapsed that into one call and one policy — and the review queue dropped by more than half in the first month.”
CR Head of Credit RiskMid-size NBFC
About

Identify. Score. Secure.

iScoreIt is a risk intelligence platform for the teams who have to say yes or no to a stranger in under a second — and then explain that answer months later.

Work with us See the products
Why we exist

Verification answers a field. Risk teams need an answer.

The gap between “this PAN is valid” and “approve this applicant for ₹1.87 lakh” is where most of the cost, most of the fraud and almost all of the manual work sits. iScoreIt is built to close it. We come out of lending, not out of a generic API business — which is why the platform is opinionated about cost-aware waterfalls, decision replay and what happens when an upstream source goes down mid-journey.

The problem

Six vendors, one decision

Risk teams were stitching together a PAN provider, a bank verification provider, a bureau, a device SDK and two spreadsheets — then reconciling the answers by hand. The decision was never the hard part. Getting trustworthy inputs to it was.

The idea

Score, don't just verify

A verification API tells you a field matched. That is not a decision. We wanted a layer that collects the signals, weighs them against a policy you control, and returns something you can act on and later defend.

The build

One contract across every check

Same auth, same error grammar, same idempotency, same webhook signature — across identity, banking, business, device, income and AML. Learn one endpoint and you have learned all of them.

The difference

Fraud intelligence from the open web

MuleNet came out of a simple observation: illegal collection operations have to publish an account to get paid. That published evidence, gathered continuously, is a fraud dataset nobody else was assembling systematically.

How we build

Four things we will not trade away.

Evidence over assertion

Every score comes with the signals that produced it. If we cannot show you why, we do not ship it.

Fail loud, not open

A missing signal is reported as missing. Your policy decides what that means — we never quietly substitute a default.

The customer holds the relationship

Where a regulated registration belongs to you, it stays with you. We are the technology layer, not a middleman on your compliance.

Boring where it matters

Auth, idempotency, versioning and audit trails are unglamorous and non-negotiable. The interesting work sits on top of them.

Built for banks, NBFCs, fintechs and marketplaces

0+

Verification APIs

<0ms

Median response

0%

Platform uptime

0key

UAT and live, one console

Come break it on your data.

The fastest way to evaluate us is a sandbox key and your own back-book sample.

Book a demo hello@iscoreit.com

Sandbox keys the same day · Data residency in India · 99.9% uptime

04Speak · new vertical

Voice agents that verify while they talk.

Twelve Indian languages. Identity confirmed mid-conversation. Recording, transcript, consent and outcome filed against the same case your policy already reads.

Hear it on your own book Get sandbox keys
agent · collections-hi-v4 00:00
OutcomePromise to pay · 14 Sep confidence 0.87
Recording · transcript · consent stored
12Languages, code-mixed
720msTurn latency
100%Calls scored, not sampled
1Audit trail, shared
Voices

Pick a voice your customers won't hang up on.

Each agent is a voice, a language and a manner. Tune the pace, the formality and how hard it pushes — then keep the same one across every campaign so the relationship stays consistent.

आपकी किस्त 5 तारीख़ से बाकी है — क्या मैं अभी भुगतान लिंक भेज दूँ?Aarohi · early-bucket reminder
हिन्दीHindi বাংলাBengali मराठीMarathi తెలుగుTelugu தமிழ்Tamil ગુજરાતીGujarati ಕನ್ನಡKannada മലയാളംMalayalam ਪੰਜਾਬੀPunjabi ଓଡ଼ିଆOdia اردوUrdu EnglishIndian English

Code-mixed Hindi–English handled natively · the agent switches when the customer switches

What makes it ours

Any vendor can place a call. Ours comes back with evidence.

Verifies mid-call

Identity is confirmed inside the conversation, not assumed from the dialled number.

Mobile to namematched
PAN last fourconfirmed
Date of birthmatched
DigiLocker linkpushed on SMS
New payout accountMuleNet check

Stops where the rules stop

Conduct rules are configuration, not training data. The agent logs the reason when it refuses.

08:00window opens 19:00closes
DND and opt-outsuppressed
Coercive languageblocked
Distress or disputehuman handoff

Acts before it hangs up

The call ends with something done, not a note for someone to action tomorrow.

Payment link on WhatsAppsent
Callback booked14 Sep
LOS and CRMupdated
Pulse policyre-run on new data
Case fileclosed

Dialect Listen — quality assurance on all of it, not two per cent of it

Sampled QA finds the calls you happened to pull. Listen scores every call your team makes, agent or human, for intent, promise-to-pay likelihood, dispute language, hardship signals and conduct breaches — so a supervisor argues with evidence rather than memory.

82/100Intent to pay
96%Conduct compliance
3Dispute flags this week
11%Contact rate lift
Promise to pay41%
Callback requested23%
Dispute or hardship9%
Not reachable27%

Figures illustrative, pending measured data from live deployments

Where teams point it

The calls nobody has enough people for.

Early-bucket collections

Reminders, promise-to-pay capture and a payment link before the account ages into a costlier queue.

day 1–30
Lead qualification

Inbound and campaign leads answered in seconds, intent captured, only the worthwhile ones routed to a human with full context.

inbound
Re-KYC and profile refresh

Periodic reviews run as a conversation, with the DigiLocker link pushed mid-call and the document fetched before the customer hangs up.

outbound
Address and delivery confirmation

Undelivered card, cheque book or device: the agent confirms the address, updates the record and reschedules in one pass.

NDR
Onboarding assist

Applications abandoned at document upload get a call that explains the missing step and completes it on the line.

recovery
Support and disputes

Balance, statement, EMI date and dispute intake around the clock, with anything sensitive handed to a person immediately.

24×7
One key, same contract

A call is just another check on the decision endpoint.

Same authentication, same idempotency, same webhook signature as every other iScoreIt API. Name an agent, name a policy, and the outcome comes back as structured data — not a recording someone has to listen to.

01
Queue and screen

Numbers checked against DND, opt-outs and the calling window before a call is placed.

0 ms
02
Connect and identify

Recording notice read, consent captured, mobile-to-name and one knowledge check confirmed.

11 s
03
Hold the conversation

Agent follows your script and guardrails, handles objections, escalates on distress.

78 s
04
Act on the outcome

Payment link sent, callback booked, systems updated, policy re-run if the picture changed.

96 s
05
File the evidence

Recording, diarised transcript, consent artefact, reason codes and agent version stored on the case.

closed
Consent captured before the conversation Recording notice in the customer's language Diarised transcript with timestamps Retention window you set Data residency in India DPDP-aligned processing Agent and prompt version pinned to every call Bring your own SIP trunk, or use ours

Give it a hundred of your own calls.

Send a bucket you already work — a delinquency slice, an abandoned-application list, a re-KYC batch — and hear the difference against your current contact rate before you commit to anything.

Book a demo hello@iscoreit.com
Sandbox keys the same day · Bring your own trunk · Data residency in India
Explore the APIs Book a demo