Electronic expense reading

In the app, the New expense card in Expenses has an Electronic expense reading tab. Drop PDFs or photos of invoices and receipts into it, several at once if you like, and we read the supplier, company ID (IČO), VAT number (DIČ), document number, dates, amounts and VAT breakdown.

1. The first time, you confirm that you understand where documents are sent to be read. Once per company.
2. Upload the files: PDF, JPG, PNG or WebP, up to 15 MB each. They are read in the background, you can close the page.
3. For a document that has been read, click Check. The expense form is pre-filled and the original is right next to it.
4. Check the details and save the expense; the document is attached to it.

While reading we check the company ID against the company register, the VAT number against the VAT payer register and the totals against the VAT rates. Anything we couldn’t read reliably is highlighted, and we warn you about a document you already have in your expenses.

VAT. For VAT payers we pre-fill a regular Czech invoice with the deduction. For reverse charge we switch it on, calculate the tax from the base and suggest the supply code. We don’t deduct foreign VAT (such as German 19 %) and we explain purchases from the EU or abroad without VAT. A document in a foreign currency is converted at the rate on its issue date.

How many documents. Free plan: 5 to try. Standard: 2 a month. Business: 200 a month. A document we failed to read because of an error on our side doesn’t count, nor does one you discard before it was sent for reading.

Where documents go. This is the only fakturum feature that sends the content of your documents to be read by artificial intelligence. The file is read by the Claude model from Anthropic, which gets only the document itself. Anthropic stores the data in the USA, doesn’t use it for training and deletes it within 30 days. You can withdraw consent at any time, and entering an expense by hand remains an equal option that sends nothing anywhere.

Invoices in ISDOC (also PDFs with ISDOC inside) are loaded without AI in the Import and other ways tab and don’t count towards the limit.