It doesn't just read bank statements. It proves them.
Upload a bank statement in any format — PDF, scanned PDF, CSV or Excel — and get back a colour-coded Excel workbook where every single transaction row is mathematically proved against the statement's own running balance, plus business categorisation and a Tally-ready import file.
Why CAs use it
Every row proved
Checked against the statement's own running balance — VERIFIED or FLAGGED with a reason, never silently dropped.
Remembers categorisation
Tag a client's payee once — every future statement for them arrives pre-categorised.
100% offline
No AI, no API key, no network call. Runs fully air-gapped — client data never leaves your machine.
Tally-ready in one click
Reconciled records export straight to Tally XML, with masters created automatically.
Categorise a client once. Never again.
Tag a counterparty as “Rent” or “Supplier Payment” on this month's statement, and StatementIQ remembers it for that client forever. Next month's statement — and every one after — arrives with that payee already categorised. Multiply that across a firm running dozens of clients a month, and the categorisation work you do shrinks to almost nothing after the first cycle.
Learned once → applied on every future statement, automatically
Runs fully offline. No client data ever leaves your machine.
There is no AI model, no API key and no network call, ever — StatementIQ runs entirely on your own laptop or firm server, even air-gapped with no internet connection at all. For a CA handling confidential client financials, that means zero exposure: nothing is ever uploaded to a third-party cloud, no data retention policy to worry about, no breach surface beyond your own machine.
100% offline · zero data egress · works air-gapped
The Differentiator
Most tools read a statement. This one proves it.
- Every row is checked: previous balance ± this row's amount = this row's balance.
- A row is marked VERIFIED only if that arithmetic holds — to the paisa.
- Anything that fails is FLAGGED with the exact reason, never silently dropped.
- Anything unreadable becomes a visible AUDIT ROW — no clean verdict until it's resolved.
- No AI model, no API key, no network calls. Same input always produces the same output.
23
Issue codes tracked
15+
Banks auto-fingerprinted
31
Built-in P&L categories
3
Tally export formats
0
Network calls at runtime
Input Formats
Any statement, one identical reconciliation pipeline
Every format is converted into the same internal structure, so a CSV and a 40-page scanned PDF go through identical reconciliation logic.
| PDF (text layer) | Native text + table extraction via pdfplumber |
| PDF (scanned / image) | Automatic per-page OCR fallback (Tesseract) |
| Password-protected PDF | Opens the file only — password is never stored, logged or written to the audit sheet |
| CSV / TSV / TXT | Delimiter and encoding auto-sniffed (UTF-8, UTF-8-BOM, CP1252, Latin-1) |
| XLSX / XLSM | One page per worksheet, formula results read |
| XLS (legacy) | Supported — HDFC's net-banking export format |
Bank Coverage
Auto-fingerprinted across 15+ banks — and layout-agnostic beyond that
Bank detection is a convenience, not a dependency. The parser is layout-agnostic, so unrecognised banks and unusual layouts still process correctly.
Extraction Engine
Maths decides the layout, not a guess
Header alias matching
Recognises every name a bank uses for the same column — Date/Txn Date/Value Date, Debit/Withdrawal/Dr, Balance/Closing Balance and more.
Arithmetic layout inference
When a table has no recognisable header, StatementIQ brute-force searches column layouts and accepts one only if it satisfies the table's own running-balance maths.
Never loses a row silently
A line that looks like a transaction but fails to parse is preserved as an audit row, never discarded — with a per-page coverage audit to catch shortfalls.
Honest OCR
A page that needs OCR but can't get it is reported as OCR_UNAVAILABLE, not returned as a blank page.
12 date formats & smart amount detection
Reference numbers are never mistaken for currency; round amounts and formula-derived cells are still recognised correctly.
Automatic metadata scrape
Bank name, account holder, account number, statement period, opening/closing balance and the statement's own printed totals.
Counterparty Intelligence
Knows who you paid, even when the bank doesn't say it clearly
- Reference-token cleanup merges references like AXOIC30738837963 into one clean party name.
- Rail-aware grouping — NEFT/RTGS keyed by name + account fragment, UPI keyed by normalised full name.
- Guarded fuzzy consolidation (Jaro-Winkler ≥ 0.92) merges bank-truncated spellings without over-merging different entities.
- Salary payer identification and Top 10 debit / Top 10 credit counterparties ranked by aggregated total.
Business Categorisation
Books-ready categories, learned once per client
- 31 built-in categories aligned to the Indian Schedule P&L expense heads — straight into the books.
- Unlimited custom categories that persist forever, behaving exactly like built-ins.
- ~25 deterministic auto-rule families tuned for Indian vocabulary: MSEDCL/Tata Power, Airtel/Jio, Ola/Uber/IRCTC, GST/TDS/EPFO, Zerodha/Groww, LIC/HDFC Ergo and more.
- 22 payment gateways detected by name — Razorpay, PayU, PhonePe, Paytm, Stripe and others.
- Learns and never asks twice — categorise a counterparty once, every future statement for that client arrives pre-filled.
- Conservative by default — anything a rule isn't confident about stays Uncategorized. Categorisation can never change a figure.
The Excel workbook
A five-sheet, colour-coded, print-ready workbook with click-through drill-down and a SHA-256 fingerprint of the source file.
- Validation Summary — Verdict, headline numbers, full provenance, Top 10 debit/credit blocks
- Pivot Summary — Every counterparty grouped, with debit/credit totals and category
- Transactions — Every row and every check — date, counterparty, category, status, issue codes
- Counterparty Detail — Per-counterparty transaction listings
- Extraction Audit — Per-page evidentiary trail — native text vs OCR, confidence, unreadable pages
Tally integration
Three export formats built from the same reconciled records — with correct voucher types (Payment / Receipt / Contra) and automatic ledger grouping.
- Tally XML — Native <ENVELOPE><IMPORTDATA> — ledger masters first, then vouchers
- CSV daybook — One line per voucher, for third-party importers
- Master Creation Sheet (.xlsx) — Ledger masters only
Privacy & Security
Your client's financial data never leaves your server
- Zero data egress — no LLM, no API key, no third-party service, no network call at runtime. Works fully air-gapped.
- Files are deleted after every run — the upload and the generated workbook are both unlinked.
- PDF passwords are never stored, logged, or written to the audit sheet.
- The only persisted file is the counterparty→category map — names only, no amounts, dates or account numbers.
- Login-gated app: server-side gate, SHA-256 password digest, brute-force lockout after 5 attempts.
Who It's For
Built for anyone who currently retypes bank statements into Excel
“Every row proved against the balance — or flagged with the exact reason.”
“It learns your clients' payees once, then never asks again.”
“Statement in. Reconciled Excel and a Tally import out.”
“No AI guessing. No API key. Nothing leaves your server.”
“The only tool that tells you what it couldn't read.”
“From PDF to Tally in one screen.”
₹8,000
per year
- Unlimited statements & banks
- Unlimited counterparties & custom categories
- Excel + Tally exports
- Self-hosted or hosted
- No per-page or per-transaction charge
Ready to Build Your AI Advantage?
Let's discuss how AI agents and automation can transform your business.