Bank statement intelligence · for lenders

Every rupee, every red flag, in sixty seconds.

Obsrv reads a bank statement the way your best credit analyst does, income, obligations, bounces, tampering, then shows the math behind every number. Analyze one statement, or consolidate a borrower's accounts into a single report. Built for NBFCs, DSAs, and lending teams. Self-serve, pay per statement.

No enterprise contract · no sales call · early lenders shape the roadmap

Statement analysis#OBV-2481
HDFC BANK · SAVINGS · XXXX4821 · DEC ’25 TO MAY ’26 · 214 TXNS
Avg monthly income₹85,000verified ✓
Avg monthly balance₹64,800
Existing obligations₹39,200 / mo
FOIR (debt-to-income)46%
NACH bounce1 · Feb 2026flagged
Balance reconciliationPASS · 214/214
VerdictMEDIUM RISK · 68 / 100

rendered from sample data · open the full report →

income verification · NACH / ECS bounces · FOIR · counterparties · cash deposits · tamper check · balance reconciliation · churn of funds

The status quo

Line by line, in Excel, after midnight.

Most lending teams still underwrite statements by hand, and the cost shows up as slow turnarounds and missed flags.

The hour per file

An analyst reads page by page, re-keys numbers into a spreadsheet, and the queue of applications keeps growing.

The missed flag

A NACH return on page 14, an EMI buried in narrations, a tampered PDF: the things a tired human read slides past.

The gated tools

Enterprise analyzers exist, behind sales calls, annual contracts, and pricing built for banks, not for your team.

What you get

The signals you actually underwrite on.

From any statement, whether digital, scanned, or password-protected, into one decision-ready report. Export PDF or JSON; API with early access.

Income & cash flow

  • Salary / income detection & regularity
  • Average & minimum monthly balance
  • Net monthly inflow vs outflow
  • Top counterparties received from
  • Negative-balance days

Obligations & capacity

  • Existing EMIs & recurring obligations
  • FOIR / debt-to-income view
  • Top counterparties paid to
  • Estimated disposable income
  • Channel mix · UPI / NEFT / RTGS / cash

Risk flags

  • Bounced payments: NACH / ECS
  • Cheque returns & penal charges
  • Circular & suspicious transfers
  • Statement tamper detection
  • Anomalies explained in plain English

Method

Three steps. Under a minute.

01

Upload

Drop in the statement: digital PDF, scanned, or password-protected. Multiple months and accounts welcome.

02

Extract & verify

AI transcribes every transaction; deterministic checks reconcile each line against the running balance. Anything that doesn’t add up is flagged, never guessed.

03

Decide

A decision-ready report plus a recommendation (approve, refer, or decline) sized to your credit policy, with the reasons behind it. Your team approves, refers, or overrides; the call stays yours.

Borrower view

Consolidate a borrower's accounts into one report.

Bring six to twelve months across every account and bank into a single report, with one FOIR computed over all of it.

All accounts in one view

Combine multiple months and banks into a single report: income, obligations, and one FOIR computed across every account, instead of reading from a single statement.

Self-transfers netted out

When a borrower moves money between their own accounts, Obsrv matches the two legs and removes them, so income and spend are not counted twice. Each one is shown with a confidence score so you can check it.

Gaps and mismatches shown

Missing months, balance jumps between statements, and statements held under a different name are flagged, so you always know what the consolidated view is built from.

Every transfer Obsrv nets out traces back to the two source rows it came from

Why you can trust it

Numbers you can audit.

In lending, a wrong number is worse than a slow one. So Obsrv is built so you never have to take the AI’s word for anything.

The math is deterministic

AI transcribes; code calculates. Every aggregate (FOIR, balances, cash flow) is computed from the extracted ledger and re-checked against the running balance, line by line.

Uncertainty is flagged

Every extracted figure carries a confidence score. Anything below the bar is routed to human review, never silently passed into your report.

Data is not the product

Statements are encrypted in transit and at rest and processed in India. By default raw files are deleted as soon as the report is produced; you can opt into a retention vault and control deletion yourself. Applicant data is never sold or used to train models.

Why teams trust Obsrv

Obsrv, as in observe.

In lending, a wrong number is worse than a slow one. Obsrv is built so you never have to take the software's word for anything: the math is deterministic, every figure is reconciled, and uncertainty is flagged, never hidden.

A registered Indian company

Obsrv Technologies LLP. A real entity accountable for your applicants' data, not an anonymous tool.

Processed in India

Statements are encrypted in transit and at rest and processed in India (AWS Mumbai), aligned with the DPDP Act. You control retention and deletion.

Engineered for verification

Built by engineers from banking technology backgrounds. Every number is reconciled against the running balance and the risk score is computed by auditable code, not guessed by a model.

Questions, security reviews, or onboarding → [email protected]

Pricing

₹5 a page.
That's the whole price list.

One page analyzed costs one credit. Buy credits, spend them on statements. No subscriptions, no seat fees, no sales calls.

Start with free credits

1 page = 1 credit = ₹5

A 12-page salary account costs ₹60 to analyze. A 60-page current account costs ₹300. CSV exports count 40 transactions as a page, so the same statement costs the same either way.

Short on credits? We say so.

If your balance covers 3 of 5 pages, we process 3, tell you exactly which pages remain, and hold the document for 7 days. Top up and resume where it stopped.

Early access starts free

Onboarded lenders get free credits: enough to run a real statement end to end before paying a rupee.

FAQ

Asked, answered.

What does Obsrv do with a bank statement?+

You upload a statement (PDF or CSV) and get an underwriting report in about a minute: verified income, existing EMIs and obligations, FOIR, NACH/ECS bounces, top counterparties, integrity checks, and a plain-English verdict your credit committee can read in one pass.

How much does it cost?+

₹5 a page: one page analyzed costs one credit. CSV exports count 40 transaction rows as a page, so the same statement costs the same either way. Credits are prepaid; there are no subscriptions, seat fees, or sales calls. New organizations start with a free trial page.

Why should a lender trust AI-extracted numbers?+

The AI only transcribes. All money math is computed deterministically, and every transaction is reconciled against the running balance, line by line. Anything that doesn't add up is flagged or held for human review, never silently passed into your report.

What happens to the statements I upload?+

Statements are encrypted in transit and at rest, processed in India (AWS Mumbai region), and raw files are deleted automatically after analysis. Applicant data is never sold or used to train models. Reports remain available to your account.

Which formats are supported?+

Digital PDFs, scanned PDFs, password-protected PDFs (provide the password at upload), and CSV exports from net-banking. Multiple banks and statement layouts are handled without per-bank templates.

Does Obsrv make the lending decision?+

No, you do. Alongside the report Obsrv gives a recommendation (approve, refer, or decline) sized to your credit policy, with the reasons and the affordability math behind it. Your underwriters approve, refer, or override it, every override is logged, and the report stays an aid to your decision, not the decision itself.

Can I analyze several statements for one borrower together?+

Yes. Group a borrower's statements into a case and Obsrv consolidates them into one reconciled report: combined income, a single FOIR across all accounts, and a merged cash-flow view. Money the borrower moves between their own accounts is detected and netted out, so income and spend are not double-counted, and every netted transfer is shown with a confidence score so you can audit it. Missing months and identity mismatches are flagged, never silently merged.

Early access

Get in before the queue does.

Early lenders are onboarded first and decide which banks and signals we build next.

  • 01Join the waitlist. Takes a minute, no commitment.
  • 02We onboard lenders in small batches and analyze your real statements with you.
  • 03You get a tool tuned to your banks, your formats, your credit policy.

questions → [email protected]

Join the waitlist

human check · … + … =

No spam. Only your onboarding invite and the occasional build update