What the Analyser Extracts
Clean Transaction List
Date, narration (cleaned), debit/credit amount, balance after transaction, transaction type (NEFT/IMPS/UPI/RTGS/NACH/ATM/CDM). UPI VPA extracted from narration.
Income Categorisation
Salary credits (identified by employer name in narration), business receipts (UPI/NEFT inflows), rental income, interest credits, GST-linked business turnover. Monthly income trend over 6 months.
EMI and NACH Detection
Standing instruction and NACH auto-debit patterns identified by narration keyword matching and recurrence analysis. Approximate existing EMI count and monthly obligation amount — critical for FOIR computation.
Risk Flag Engine
Cheque bounce / return transactions, overdraft utilisation frequency, cash deposit concentration (potential income inflation), round-trip transfers, irregular salary credit pattern, end-of-month balance trend.
Underwriting Summary
Average monthly balance (AMB), minimum balance, net monthly inflow (NMI), total credits vs. debits, salary-to-obligation ratio, credit utilisation. Underwriting decision ready in <3 seconds after PDF upload.
50+ Indian Banks Covered
SBI, HDFC, ICICI, Axis, Kotak, PNB, BOI, Canara, Union Bank, Bank of Baroda, IDFC First, YES Bank, IndusInd, Federal, RBL, IDBI, DBS India, Bandhan, and 30+ more cooperative and regional banks.
Sample API Output
"account_holder": "RAJESH SHARMA",
"bank": "HDFC Bank",
"period": { "from": "2025-10-01", "to": "2026-03-31" },
"summary": {
"avg_monthly_balance": 48250,
"net_monthly_inflow": 85000,
"total_emi_obligation": 22000,
"emi_count": 2,
"foir": 0.26
},
"income": [
{ "type": "salary", "employer": "TATA MOTORS LTD", "avg_monthly": 72000 },
{ "type": "freelance", "avg_monthly": 13000 }
],
"risk_flags": [
{ "flag": "cheque_bounce", "count": 1, "severity": "low" }
],
"transactions": [ "..." ]
}
Supported Input Formats
Use Cases
Loan Underwriting Automation
Automated income verification, FOIR computation, existing obligation detection. Reduce underwriting TAT from 2 days to 5 minutes. Integrates with your LOS (loan origination system) via REST API.
Client Bookkeeping from Bank Data
CA firms process hundreds of client bank statements monthly. AI extraction auto-categorises transactions into income/expense heads, matches GST-liable entries, and exports to Tally or accounting software.
Credit Score Enhancement
Supplement bureau scores with cash flow analysis. Borrowers with thin bureau files but strong bank statement history can be underwritten. Particularly valuable for new-to-credit MSME and self-employed borrowers.
AML / KYC Transaction Monitoring
Flag structuring patterns (cash deposits just below ₹10L threshold), suspicious round-trips, and unexplained large credits. Designed for NBFC compliance teams and RBI reporting requirements.
Pricing
- 50 bank PDF formats
- Transaction extraction + JSON
- Income/expense categorisation
- REST API deployment
- Private cloud (AWS/Azure)
- Everything in Basic
- EMI obligation detection
- Risk flag engine (8 flags)
- AA framework integration
- LOS API integration
- White-label report PDF
- Everything in Underwriting
- Bureau + statement fusion
- Custom ML scoring model
- Multi-lender white-label
- Real-time monitoring dashboard
- AML / transaction monitoring
vs. Perfios API: ₹3–₹15/statement at scale. At 10,000/month = ₹30K–₹1.5L/month ongoing. Custom system pays for itself at volume and gives you full data ownership.
FAQ
Can the system handle password-protected bank statement PDFs?
Yes. Most Indian bank PDFs are password-protected with the customer's date of birth or account number. The system accepts the password as a parameter alongside the PDF. For NBFC loan processing, the customer provides the password during the consent/document upload step. We also support cases where customers forward PDFs from WhatsApp directly to an API endpoint.
How is this different from Perfios or Karza?
Perfios and Karza are multi-tenant SaaS APIs — your data goes through their shared infrastructure, you pay per call forever, and you can't modify the categorisation taxonomy. XPndAI builds a private-deployed system on your own AWS or Azure account. Your data never leaves your infrastructure, you own the categorisation rules and risk flag thresholds, and there are no per-call fees after build. Better fit if: you process high volumes (1,000+/month), have custom underwriting criteria, or need to maintain RBI data localisation compliance with full audit logs.
Does it support the Account Aggregator (AA) framework?
Yes. The AA framework (RBI-regulated) delivers structured JSON transaction data directly from the bank to your system without PDF parsing. We build the AA integration layer — your system registers as an Financial Information User (FIU), and the customer consents to share data via their bank's AA-enabled app. We handle both AA JSON ingestion and legacy PDF parsing so your underwriting pipeline works even for banks not yet live on AA.
How accurate is the income categorisation?
95%+ accuracy on salary detection (employer name extraction from NEFT/NACH narrations). 88–92% on business income classification (distinguishing business UPI receipts from personal transfers). Accuracy improves with calibration on your specific borrower segment — self-employed, MSME, gig workers, and salaried all have distinct narration patterns that we fine-tune to in the first month of production use.
Get a Bank Statement Analyser Demo
Share a sample bank statement (we'll redact real data) and we'll run a live demo showing extracted transactions, income categorisation, EMI detection, and risk flags.
Related: AI for CA & Accounting Firms · FinTech AI Development · AI Agent Development India