Tutorial
Create an Excel spreadsheet from PDF invoices
Ancilo can read PDF files and photos of receipts and create an Excel spreadsheet from their contents. Tasks is the appropriate area for this: Ancilo works in a copy, and only what you keep reaches your folder. This tutorial describes a suggested workflow, not a recorded test run.
Prepare the files and model
Start with a separate folder containing a few invoices stored fully on your computer. Ancilo does not read files available only in iCloud. A local model needs to be configured and fit the available memory. Tasks always use the local model: the invoices are not sent to a cloud service, and the task agent does not run commands.
PDFs with embedded text can be read directly. Since version 0.3.0, Ancilo reads scanned PDFs and photos of receipts (JPEG, PNG, HEIC, TIFF, WebP) with macOS text recognition – up to 50 pages per scanned file. Text recognized this way is marked “text recognized – may have mistakes” in the app. Blurred digits, decimal marks and similar characters can be misread, so the original invoices remain the reference for checking the result.
Specify the columns and missing values
In Tasks, click “New task” and first choose the prepared folder; the dialog opens in Documents. The folder’s size does not matter. Your home folder itself, Library and system folders cannot be chosen. You can also drag individual invoices from elsewhere onto the input; if the folder already contains the same file, Ancilo does not create a second copy.
Then enter the task. Specify the output filename, columns and how to handle missing information. Including a source for each row makes the later comparison with the invoice easier. Amounts are stored as numbers in the spreadsheet only when written without thousands separators or currency symbols – “1234.50” becomes a number, “1,234.50 €” stays text.
Read the PDF invoices in this folder. Create invoices.xlsx with one row per invoice and columns for source file, page, invoice number, date, supplier, net amount, tax amount, total amount and currency. Include only information explicitly present in the documents. Leave missing or illegible values empty and describe them in an additional Review note column. Write amounts without thousands separators or currency symbols, for example 1234.50; put the currency in the currency column. Do not change or delete source files.
Check the spreadsheet against the invoices
While it works, Ancilo shows what it is doing; the expert view also lists the individual steps. When the task ends, the change overview lists new, changed, moved and deleted files. For this task, the new Excel file should appear there; unexpected changes to source files are a reason to drop the result or request a correction.
Compare invoice numbers, decimal places, currencies and the assignment of net and total amounts with the invoices. Starting with a small set makes it possible to check every row. An empty review note does not establish that a row is correct, and a cell stored as a number can still hold a misread value.
Keep, drop or undo
Keep writes the result to the selected folder; only then does invoices.xlsx appear there. Ancilo first checks whether affected files have changed since the task began and writes nothing if there is a conflict. Drop leaves the folder untouched. After keeping, Undo remains available as long as the affected files have not subsequently changed.
This workflow does not require web search. The Web search switch next to the input applies to chats, tasks and coding alike. When it is on, Ancilo searches without asking, and search terms could contain information from the invoices. Turn the switch off for this task: Ancilo then asks before every search, and you can decline.