About FlowParse

FlowParse is built and run by Kyrylo Mikushyn, its founder, through a company registered in Denmark under CVR number 46634675, based in Silkeborg. There is no holding structure, no call centre and no separate sales team. The person who answers your email is close to the person who fixes the bug you found, and often the same one.

That shapes how the product is made. Every claim on this site has to be something we can stand behind, because there is no one else to blame when it turns out wrong. When a customer reports a statement that came out wrong, it is read, reproduced and fixed, and the fix is checked against the other documents we hold, so one customer's problem does not quietly become another's.

You upload a PDF, a scan or a phone photo of a financial document: a bank or card statement, an invoice, a receipt, a payroll register, a brokerage or loan statement, a financial report. FlowParse works out what kind of document it is and reads it into a table. No templates to set up, and it does not need to know your bank in advance.

The result downloads as Excel, CSV or Google Sheets, or as an import file for QuickBooks (QBO), Quicken (QFX), OFX, MT940, Xero, Sage, DATEV, Zoho Books, NetSuite, MYOB, Wave, FreshBooks, FreeAgent, Tally and 1C. Smart Merge combines up to 100 statements into one workbook, and invoice matching shows which invoices a bank statement has paid. Most documents are ready in under a minute. A REST API covers the same steps for developers.

A bank statement carries its own proof. The opening balance plus every transaction must equal the closing balance the bank printed, and on most statements every row's running balance must follow from the row before it. FlowParse runs that test on every statement it reads. When the arithmetic does not hold, the gap is flagged and shown to you before export. It is never smoothed over, and a statement that fails its own sum is never presented as clean.

Invoices get the same treatment with their own arithmetic: line items against the subtotal, subtotal plus VAT against the total, the VAT rate against the amounts, and the VAT number against its country's format. You can correct any cell in the preview, the checks run again, and the file you download contains your corrections.

We think this is the honest way to automate bookkeeping input. Reading a document is the easy half. Knowing whether the reading is right is the half that decides whether you can trust the spreadsheet.

Files travel over an encrypted connection and our servers are in the EU. The text and page images are sent to our processing providers only to read the document, and some of them process data outside the EU. The Security page and Privacy Policy list the details.

The original file is kept for 30 days, so you can compare the result with the source and a failed read can be retried, and is then deleted automatically. The extracted data stays in your history until you delete it, and each document has a delete button that removes the data and the original together. Your documents are never used to train AI models and are never sold.

Being clear about the edges saves you time, so here they are.

The free plan includes 10 pages a month with no card, so you can try FlowParse on your own documents first. Paid plans start at €15 a month for 300 pages. A page is a page of the document, and pages are only counted when a document is read successfully. A failed read costs nothing. Every plan, including the free one, includes every export format.

Documents vary endlessly, and sometimes a reading is wrong. When that happens, the checks are meant to tell you before you rely on the file. If they miss something, write to us with the document name and what you expected. Real reports are how the product gets better, and we would rather hear about a problem than have you work around it.