Finance & admin

Invoice to Excel converter

Get invoice tables: line item, SKU, quantity, unit price, tax, discount, subtotal, total, vendor text, and invoice date when visible.

Scans detected automatically · AI Deep Read only when needed
Free: 5 conversions a day · 5 MB each
Deleted after 24 h · never used for training

Before you start

What to expect

Files are retained only for the product retention window and deleted after 24 hours unless your account policy says otherwise.
Uploads are not used to train models; AI Deep Read reads the page only to produce your requested output.
AI Deep Read costs 1 credit per rendered page and still needs human review for financial records.
We extract readable invoice tables, not guaranteed AP automation fields; verify vendor, invoice number, tax, and totals manually.

What the output contains

Line-item tables read from invoices, with numeric quantities, prices, tax, and totals typed as spreadsheet numbers when recognizable.
Header-like details such as vendor, invoice number, date, currency, and due date only when they are visible in or near the table output.
Separate sheets or table blocks for multi-invoice PDFs when layouts differ too much to stitch safely.
JSON or CSV can be selected after review, but the workflow starts with XLSX because accountants can inspect it quickly.

A real run

Contractor invoice packet

Input: 6-page PDF containing three invoices, each with 8 to 14 service rows, mixed USD and CAD symbols, and tax lines below the table.
Output: Workbook with three sheets, 34 line-item rows, typed Qty/Unit price/Tax/Total columns, and two vendor-address rows that needed moving out of the table before import.

Known failure modes and how they're handled

Line items mixed with header fields: The workflow prioritizes tables the model reads. Header fields may appear as nearby rows, but there is no dedicated invoice field-extraction model.
Multi-invoice PDFs: AI Deep Read processes each rendered page and separates obvious table regions; review invoice boundaries before bulk posting.
Currencies, VAT, and locale formats: Recognizable currency symbols and decimal formats are preserved or typed, but mixed locales should be checked before formulas are added.
Photos with shadows or tilted paper: Images route to AI Deep Read, which can read many photographed tables, but glare and cropped totals can produce missing cells.

Command line

Do it yourself

tabula-py

bash

python -c 'import tabula; tabula.convert_into("invoices.pdf", "invoice-lines.csv", pages="all", lattice=True)'

Useful for ruled digital invoices; scanned photos still need OCR and custom cleanup.

camelot

bash

camelot -p all -f xlsx -o invoice-lines.xlsx invoices.pdf

May miss borderless tables and will not reliably distinguish invoice header fields from line items.

Questions about this workflow

Does this extract vendor, invoice number, and total as fields?

It may include them when visible, but the promised output is the table or tables AI Deep Read reads. It is not a dedicated invoice field-extraction API.

Can I upload a photo of a paper invoice?

Yes. JPG, PNG, and HEIC invoice photos use AI Deep Read. Crop and brighten the image first for better line-item accuracy.

How are multi-page invoice tables handled?

Repeated headers are used to stitch continuing tables where possible. Different invoice layouts may become separate sheets or blocks.

Can I export to JSON for an accounting integration?

The workflow starts with XLSX for review, and table outputs can be converted to CSV or JSON after you verify columns and totals.

← All workflows