Tutorial
10 min

Process Invoices

Upload CSV or PDF invoices, use the field inspector to identify key columns, and let Claude extract and summarize line items into a clean, exportable format.

Prerequisites

  • A Workhog account (free or paid)
  • One or more invoice files in CSV or PDF format
  • Basic familiarity with the Workhog chat interface
Example

Example invoice files:

  • acme-invoices-jan-2026.csv
  • vendor-billing-q4.pdf
  • supplier-statement-dec.xlsx

Sign in to see your actual files here.

Step-by-Step

1

Upload your invoices

Drag your invoice files onto the sidebar, or click the + icon in the sidebar header and select Upload files. You can upload multiple files at once.

Supported formats: .csv, .xlsx, .xls, .tsv, and .pdf.

Organize first

Create an invoices/ folder in the sidebar before uploading. This keeps your workspace tidy and makes it easier for Claude to find related files later.

2

Open in the viewer

Click a file in the sidebar to open it in the document viewer. CSV files open in the spreadsheet viewer with sortable columns. PDF invoices open in the PDF viewer with page navigation.

Take a moment to verify the file loaded correctly and review its structure.

3

Use the CSV inspector to select key columns

For CSV/Excel files, click any cell in the spreadsheet viewer. The Field Inspector panel opens, showing:

  • The column name and data type (text, number, date, currency)
  • A sample of values from that column
  • Basic statistics (count, unique values, min/max for numbers)

Click the columns that contain invoice numbers, dates, amounts, and vendor names. Each click sends context to Claude so it knows exactly what your data looks like.

Info

The Field Inspector is much more precise than describing your columns in chat. It sends the exact column metadata, data type, and sample values to Claude.

4

Tell Claude to extract line items

With the inspector context loaded, ask Claude to process your invoices:

Example prompttext
Extract all line items from this invoice CSV. For each item, pull out the invoice number, date, vendor name, description, quantity, unit price, and total. Flag any rows where the total doesn't match quantity x unit price.

Claude reads the file from your workspace, uses the field inspector context you provided, and extracts structured data. You'll see tool calls appear inline as it processes each section.

5

Review the extracted data

Claude presents the extracted line items in a formatted table. Review the results for:

  • Missing or mismatched fields
  • Flagged rows with calculation discrepancies
  • Any vendor names that need standardization

If something looks wrong, tell Claude what to fix. It can re-read specific rows or adjust its extraction logic.

6

Export as a summary CSV

Once you're satisfied with the extracted data, ask Claude to save it:

Example prompttext
Save the extracted line items as a CSV file called invoice-summary.csv with columns: Invoice #, Date, Vendor, Description, Quantity, Unit Price, Total, Status

The file appears in your sidebar. Click it to open in the spreadsheet viewer and verify the output. You can download it from the viewer toolbar or share it via Google Sheets.

Batch processing

Have dozens of invoice PDFs? Use the PDF Batch Processor app instead of processing them one at a time. Upload all your PDFs, select the data fields to extract, and Claude processes them in parallel.

Next Steps

Tutorial: Process Invoices - Workhog Help