Parse any Indian bank statement — HDFC, SBI, ICICI, Axis, Kotak, Yes Bank, PNB, and 35+ more — into structured JSON with every transaction, income classification, EMI detection, and cash flow summary. Built for NBFCs, credit teams, and account aggregator pipelines.
POST any document. Get structured JSON back. Confidence scores per field. Integrate in an afternoon.
// POST /v1/extract — bank statement { "document_type": "bank_statement", "file": "sbi_statement_q2.pdf", "options": { "classify_transactions": true, "detect_salary": true, "detect_emi": true } }
{
"account_holder": "Priya Sharma",
"account_number": "XXXX7832",
"bank": "State Bank of India",
"period": { "from": "2026-04-01", "to": "2026-06-30" },
"avg_balance": 41800,
"monthly_credits": [128400, 131200, 129600],
"salary_detected": { "amount": 72000, "day": 1 },
"emi_obligations": [{ "amount": 18500, "label": "Auto Loan EMI" }],
"transaction_count": 89,
"confidence": 0.96
}Feed structured bank statement JSON directly into your underwriting model. Income, EMI obligations, avg balance, and cash flow volatility — all structured, ready for credit scoring.
Monthly credit/debit totals, income stability score, obligation-to-income ratio — all computed. Ready for FOIR calculation, repayment capacity analysis, and risk flags.
Flag suspicious patterns: sudden large deposits before loan application, circular transfers, same-employer-same-company salary patterns, low-balance spikes. Delivered as risk signals.
For applicants without AA consent, PDF bank statement parsing delivers the same structured data. Same JSON schema as AA output for uniform downstream processing.
DSAs upload bank statement PDFs through your portal. Extract API processes them, validates completeness, and routes structured data to the originating lender API.
XPndAI Extract + XPndAI Underwriting Agent: bank statement in → credit decision out. Document extraction as infrastructure for AI agent workflows.
Start with 50 free test pages. Pay per page after that — no subscription required at low volumes. Enterprise pricing for 1,000+ documents/month.
XPndAI Extract supports 40+ Indian bank formats including all major banks (HDFC, SBI, ICICI, Axis, Kotak, Yes Bank, PNB, Bank of Baroda, IndusInd, IDFC First, AU Small Finance) and regional banks. Both Net Banking PDF downloads and scanned/photographed statements are supported. Password-protected bank statement PDFs can be handled with the password passed as a parameter. New bank formats can typically be added in 3–5 days.
XPndAI Extract uses a combination of transaction amount pattern recognition, credit timing analysis (salary credits typically on 1st–5th of month), and description keyword matching (NEFT, RTGS, salary codes, employer names). EMI detection uses debit amount regularity, NACH mandate patterns, and lender name matching. Confidence scores are provided per detected income/EMI entry so your credit team can review flagged items.
Yes. XPndAI Extract outputs bank statement data in a JSON schema compatible with the RBI Account Aggregator framework's financial information format. This means the same downstream credit processing logic can handle both AA-sourced data and PDF-extracted data without separate handling. Migration from PDF-only to AA + PDF fallback is straightforward.
XPndAI Extract achieves 97%+ transaction extraction accuracy for PDF-generated bank statements from major banks and 90–95% for scanned/photographed statements. Header fields (account holder name, account number, period) achieve 99%+ accuracy on clean PDFs. Income/EMI classification accuracy is 93%+ across supported bank formats. Confidence scores are provided at the transaction and field level for downstream quality control.
Upload any Indian bank statement PDF. Get structured JSON back — transactions, income, EMIs, cash flow. Free test, no credit card.
📋 Free API Access 💬 WhatsApp Now