Document-aware extraction
The service identifies statement metadata, table headers, transaction rows, money direction, dates, and available balance anchors. It is designed for bank statements rather than arbitrary PDF tables.
Bank statement data extraction
Extract bank statement transactions with checks you can see
Tennanova turns semi-structured PDF statements into a consistent transaction dataset. It combines document reading with deterministic normalization and balance checks, then keeps you in control through review and editing.
- Structured transaction records
- Confidence and review signals
- Deterministic balance checks
- CSV and XLSX output
- Files automatically deleted
- CSV + Excel
A transparent extraction pipeline for messy financial documents
Bank statements differ by institution, account type, period, and layout. Tennanova separates reading from validation so a plausible-looking table is not automatically treated as a perfect result.
Normalization in code
Dates and monetary values are parsed into consistent data types. Credits, debits, signed amounts, currency, and row balances are represented explicitly so downstream spreadsheets do not depend on visual spacing.
Arithmetic validation
Programmatic checks compare transaction movement with opening, closing, and available running balances. The result records whether it reconciles, needs review, or lacks enough balance information for a check.
Workflow
Read, normalize, verify, then export
- 01
Read the statement
Tennanova sends the PDF to its OCR provider for document reading. The source file stays in private temporary storage and is deleted immediately after this processing step.
- 02
Build and check the dataset
The response is normalized into a transaction schema. Tennanova calculates totals, checks balance anchors, and identifies rows that deserve review before the data is used elsewhere.
- 03
Review and export
The authorized result is shown as an editable preview. Once reviewed, export it to CSV for portability or Excel for typed cells, filtering, formatting, and a separate extraction summary.
Answers
Frequently asked questions
What data can Tennanova extract from a bank statement?
Tennanova extracts statement metadata and transaction fields when they are present and readable. These can include institution, masked account details, statement period, currency, opening and closing balances, dates, descriptions, debits, credits, amounts, and running balances.
How does balance reconciliation improve extraction quality?
Reconciliation compares the extracted opening balance, credits, debits, and closing balance. A zero difference is useful evidence that the movement was captured consistently. It cannot prove that every description or date is correct, so the preview still matters.
What happens when a row looks uncertain?
Tennanova preserves confidence and review signals with the result. Rows that conflict with available running balances can be highlighted, and you can inspect and edit supported fields before export.
Can Tennanova process combined statements for multiple accounts?
No. Tennanova verifies one account balance at a time. For a portfolio or combined document, download a separate statement for each account and convert the files individually.
Is extracted bank data used to train advertising systems?
No. Statement contents, filenames, balances, account identifiers, and transaction data are excluded from Tennanova product analytics and the Reddit Pixel. OCR providers receive statement content only to perform the requested conversion, as described in the privacy policy.