The bank statement analyser built for Indian credit teams.

It converts bank statements into a credit file. Statements arrive as native e-PDFs, scans or Account Aggregator JSON, in hundreds of layouts that change without notice. An analyser extracts every transaction from those layouts, removes noise, classifies each entry, and computes the metrics an underwriter would otherwise assemble by hand in a spreadsheet.

Fiscus runs every case through six stages:

• Intake: native PDF, scanned PDF or AA JSON, multi bank and multi account, up to 24 months deep.

• Verification: a tampering verdict on every document before a single number is used. Metadata anomalies, font breaks and balance trail failures are flagged with the exact page.

• Filtration: duplicates, reversals, and transfers between the borrower own accounts netted out so they never inflate revenue.

• Categorisation: every transaction classified with logic tuned per borrower segment, at 95%+ accuracy.

• Analysis: income, obligations, balances, counterparties and conduct computed across all accounts as one picture.

• Surface: a browser dashboard for analysts, Excel and JSON outputs, and REST APIs for your LOS.

The output is structured the way underwriters actually read a case, not as a raw transaction dump. The core signal groups:

• Income and FOIR. Income categorised across salary, bonus, EPFO, gratuity and other recurring sources for salaried borrowers, and across trade inflows for businesses. Dynamic FOIR is computed three ways: debt based, including fixed expenses, and including essentials, so policy teams can pick the lens that matches their credit policy.

• EMI and obligation conduct. Recurring EMI outflows auto tagged across all accounts, with missed and delayed payments flagged month by month and cross checked against bureau data. Obligations that never reached the bureau, including BNPL repayments and irregular app loan debits, are surfaced from the banking itself.

• EOD balances and bounces. Daily and monthly balance trends, average EOD balance, zero balance days, utilisation and overdrawn days per account. Inward and outward bounces tracked with reason codes, split technical versus non technical.

• Counterparty intelligence. Top buyers and suppliers by throughput, transaction count and average value, at account and consolidated level, with concentration measured so a single buyer dependency is visible before sanction.

• Self and related party transfers. Credits cycling between the borrower own accounts are netted out, and group, promoter and family flows are tagged using public source enrichment, so the underwriter sees real cash generation rather than a managed group view.

Every document passes fraud check units before analysis begins: template tamper validation, suspicious transaction rules and irregular pattern detection, each tuned to the borrower segment. Flags carry the exact page and rule that raised them, so review takes seconds rather than a fresh read of the file. The full detection stack has its own page: bank statement fraud detection.

Over 1000 Indian banks, including the co-operative and regional banks where formats are hardest, across savings, current, OD and CC accounts. Input can be native e-PDF, scanned PDF or Account Aggregator JSON, ingested natively.

Analysis is segment specific rather than one retail model stretched across every borrower. Templates, categorisation logic and fraud rules are configured separately for salaried individuals, self employed and micro businesses, and SME and corporate borrowers, and report fields can be configured to match your institution credit policy.

Extraction is where most tools stop and where credit work actually begins. Fiscus output is CAM ready: income, obligations, FOIR and counterparty analysis arrive structured for direct credit appraisal memo use, without rework. Obligations are cross checked against bureau tradelines, and a companion GST view benchmarks declared turnover against banked credits for business borrowers.

After sanction, the same data feeds loan monitoring and collections, and credit teams can question any case in plain language through Fiscus Copilot, with every answer citing its source page.

The browser dashboard needs no integration at all: onboarding within 48 hours, same analyst ready reports. For systems integration, unified REST APIs and webhooks for event based flows connect Fiscus to your LOS or LMS in up to two weeks, with outputs in Excel or JSON. Evaluation is designed to be zero disruption: run Fiscus in parallel with your current provider on live cases and compare.

Per statement pricing that scales with volume, designed to come in lower than traditional analysis at scale. No platform lock in to start: begin on the dashboard, integrate when the numbers have earned it.