Sprint 9 Complete  ·  Production Ready  ·  Finance Buddha

The Credit Analyst
You Can't Afford
To Hire

BankLens AI ingests bank statements, computes 220+ financial signals, runs ML-powered risk scoring, and outputs a structured credit decision — in under 5 minutes. Built for India's NBFCs, DSAs, and Fintech lenders.

0+
Financial Signals
<0 min
Processing SLA
0%
Classification Accuracy
0
Fraud Signals
banklens.vyaptix.com/applications/[id]/decision
Credit Decision
Rajesh Kumar — ₹5,00,000 Personal Loan
✓ APPROVETier B
Credit Score
74
/ 100
Avg Monthly Inflow
₹5.96L
Stable
DTI Ratio
0.0%
Excellent
Fraud Signals
0
Clean
Monthly Cash Flow
CreditsDebits
JanFebMarAprMayJunJulAugSepOctNovDec
74
Credit Score
APPROVE
ABCD
Top Reason Codes
avg_monthly_inflow+20
bounce_count+15
dti_ratio+20
salary_regularity0
Transaction Explorer
View all 127 →
SALARY CREDIT JAN 2026 — VYAPTIX TECHNOLOGIES
Salary
+₹72,500
NEFT - HOME LOAN EMI HDFC BANK
EMI
-₹28,000
UPI SWIGGY FOOD DELIVERY
Food
-₹1,850

Built for India's lending ecosystem

Finance BuddhaVyaptix TechnologiesNBFCs Tier 2–3DSA NetworksFintech LendersCooperative Banks

How it works

Statement to decision
in three steps

01

Upload Statement

Drag-and-drop any bank statement — PDF (native, scanned, password-protected) or image. Supports top 40 Indian banks with dedicated parsers plus universal OCR fallback.

02

AI Analyses 220+ Signals

A 7-stage async pipeline extracts, OCR-processes, categorises via ML, and engineers 220+ financial features — income stability, EMI obligations, fraud signals, cash flow patterns.

03

Get Instant Decision

A 0–100 credit score, risk tier (A/B/C/D), APPROVE / REVIEW / REJECT recommendation, SHAP-based reason codes, 14-signal fraud report, and a downloadable CAM PDF.

Platform capabilities

Everything a credit team needs,
nothing it doesn't

Not a report tool. A decisioning engine. One platform for ingestion, analysis, scoring, fraud detection, and compliance.

Universal Document Ingestion

PDF, scanned images, password-protected statements, Account Aggregator JSON. 40+ Indian bank parsers (HDFC, ICICI, SBI, Kotak, Axis, PNB...) plus universal OCR fallback.

220+ Feature Engineering

Income stability, EMI obligations, cash flow seasonality, salary regularity (CV score), DTI ratio, overdraft frequency, balance trajectory — all computed and stored in a feature vector.

AI-Powered Credit Scoring

0–100 credit score with risk tier (A/B/C/D) and APPROVE / REVIEW / REJECT decision. 7-rule deterministic engine today; XGBoost ML model per portfolio in Phase 2.

14-Signal Fraud Detection

PDF integrity checks, round-number transaction patterns, salary velocity anomalies, overnight crediting, ghost employer signals, duplicate narrations — all auto-scored.

Explainable Decisions (SHAP)

Every decision backed by top positive and negative score contributors with plain-English reason codes. Full rule engine pass/fail table with thresholds vs actual values.

CAM Report Generation

One-click downloadable reports: Credit Assessment Memo (PDF), 1-page summary, Excel workbook with 220+ features, and machine-readable JSON — all RBI audit-trail ready.

Role-Based Access Control

Three roles — ANALYST, CREDIT_MANAGER, ADMIN — each with scoped views and actions. Immutable SHA-256 audit log on every decision event.

Multi-Bank Statement Fusion

Upload savings + current account statements from different banks for the same applicant. Application-level deduplication prevents double-counting overlapping transactions.

Real-Time Pipeline Status

Live 7-step processing tracker — UPLOAD → OCR → NLP → CATEGORISE → FEATURES → RULES → FRAUD — with per-stage progress, error recovery, and re-processing on demand.

9 Sprints · 100% Complete · Live in Production

Built. Not planned.

Every feature below is live, tested, and running. This isn't a roadmap — it's a production platform.

Ingestion Pipeline

  • 7-stage async job pipeline with Redis distributed locking
  • pdfplumber native parsing + pytesseract OCR with OpenCV preprocessing
  • Password-protected PDF unlocking via pikepdf
  • PNG / JPG image statement support (direct OCR)
  • Multi-bank regex parser + balance-inference direction (Kotak, HDFC, SBI...)

Transaction Intelligence

  • ML transaction categorisation (SALARY, EMI, UTILITIES, FOOD, TRANSFER...)
  • Salary regularity score via coefficient of variation
  • DTI ratio from EMI-tagged transaction aggregation
  • Bounce & overdraft signal detection from narration text
  • Application-level dedup (cross-statement content hash)

Credit Decision Engine

  • 7-rule deterministic scoring engine outputting 0–100 points
  • APPROVE / REVIEW / REJECT outcomes with tier (A / B / C / D)
  • SHAP-style reason codes with contribution scores + positive/negative flags
  • Manual override with full audit trail and role-gated justification
  • Application lifecycle: DRAFT → SUBMITTED → PROCESSING → DECISIONED

Fraud Detection

  • 14 fraud signal evaluators (round-number, velocity, overnight, ghost employer...)
  • Per-signal severity: HIGH / MED / LOW with confidence scoring
  • Transaction-level is_flagged stored in DB per row
  • Fraud signal summary panel on decision dashboard

Decision Cockpit UI

  • Credit score gauge, KPI cards, reason code bars
  • Rule outcome pass/fail table with thresholds vs actual values
  • Fraud signal cards with severity indicators and transaction counts
  • Manual override modal with role-gated access (CREDIT_MANAGER+)
  • Transaction explorer: search, category filter, pagination (50/page)

Platform & Compliance

  • Multi-role RBAC (ANALYST / CREDIT_MANAGER / ADMIN)
  • Immutable SHA-256 audit log on every action
  • QStash job queue + Upstash Redis + PostgreSQL + pgvector
  • GCS file storage with signed URL upload + dev fake-GCS
  • DPDP Act 2023 consent capture on every application

The opportunity

$0B+
India digital lending market (2024)
0%
TAT reduction: 3.5h → under 30 min
0+
Registered DSA agents in India
0+
NBFCs in India needing automation

Finance Buddha — Real Numbers

₹20L saved annually.
Payback in under a month.

Finance Buddha previously spent 525 analyst hours per month at ₹400/hr reviewing statements manually — ₹2.1L/month, every month. With BankLens AI, that drops to 60 hours + ₹18,000/month subscription.

Before BankLens
₹2.1L
per month · manual review
After BankLens
₹42K
per month · 80% automated
Monthly Savings
₹1.68L
direct cost reduction
5-Year NPV
₹6.5L+
net present value

Transparent pricing

Start small. Scale fast.

No setup fees. No per-seat pricing. Pay for what you process.

Starter

₹12/ report

Pay-as-you-go for individual DSAs, CAs, loan agents, and audit firms.

  • PDF & image statement upload
  • All 220+ financial features
  • Credit score + tier + decision
  • 14-signal fraud detection
  • Summary report download
  • Up to 50 reports/month
  • DPDP 2023 compliant
Get Started Free
Most Popular

Professional

₹1.5L/ year

Unlimited reports for NBFC credit teams, DSA networks, and fintech lenders.

  • Everything in Starter
  • 500 reports/month included
  • CAM PDF report generation
  • Excel workbook export (220+ features)
  • Multi-user access (up to 10 seats)
  • REST API (1,000 calls/month)
  • Immutable audit trail
  • Priority email support
Start Free Trial

Enterprise

Custom/ year

White-label, custom ML model trained on your portfolio, and dedicated SLA.

  • Everything in Professional
  • Unlimited reports
  • Custom XGBoost ML model training
  • White-label reports + custom branding
  • Account Aggregator integration
  • On-premise deployment option
  • Rule engine admin UI
  • 99.9% SLA + dedicated support
Contact Sales

All plans include DPDP 2023 compliance · Immutable audit log · Role-based access control · No setup fee

₹1 Cr ARR Target · 25+ Clients · 5,000+ Reports/month

Cut your TAT
by 80%

Join Finance Buddha and the growing network of NBFCs and DSAs using BankLens AI to underwrite faster, smarter, and with full auditability.

No login required in dev mode · Role switcher in top bar · No credit card needed