Statement conversion data
- The statement PDF or pasted statement text
- Extracted OCR and text-layer content
- Transaction descriptions, amounts, and balances
- Edits made in the review workspace
- Generated CSV, XLSX, JSON, QBO, OFX, and QIF contents
PDF parsing, local OCR, transaction editing, and export generation happen in your browser. Account, billing, and optional analytics use separate network requests without statement contents.
Browser-based does not mean the entire website is offline. It means the sensitive conversion payload is kept separate from the services used to deliver and operate the product.
Your browser downloads the application code and may fetch PDF or OCR worker assets needed to run the converter.
Google sign-in and plan status use network requests. Statement contents are not attached to the account.
Stripe checkout, subscription status, and the billing portal are networked separately from conversion data. Job Pass limits use an opaque, atomic entitlement record with no statement contents.
When product analytics are enabled, they can record product events without statement text, transaction rows, balances, or file contents.
StatementForge describes the architecture it can support today. It does not claim bank affiliation, universal statement compatibility, or a certification that has not been independently earned.
The backend is not part of the statement parsing or export path.
Extracted rows remain visible and editable before a financial file is generated.
Account and subscription records do not need statement contents.
OCR, PDF layouts, and destination import rules can produce results that need correction.
Local processing reduces server exposure, but it cannot secure an infected device, an untrusted browser extension, or an export shared to the wrong destination.
Review the rows before saving or importing the generated file.