Manually typing invoice data into a spreadsheet is one of the most repetitive tasks in finance. Gemini can read it for you.
Every finance team deals with the same pile of paperwork: supplier invoices, expense receipts, purchase orders — each with a different layout, but all containing the same core information. Someone still has to read each document and type the supplier name, date, amount and reference number into a spreadsheet.
Gemini, used inside Google Drive and Sheets, can read an invoice or receipt — whether it is a PDF, a scanned image, or a photo — and extract the key fields directly into a structured format. It does not replace your accounting software, but it removes the slow, manual first step of getting data out of a document and into a row.
Key insight: Gemini can read the layout of an invoice — even an unfamiliar one — and map it to the fields you need: supplier, date, invoice number, line items, tax and total. You describe the structure you want; Gemini does the reading.
Extract invoice fields
Supplier, date, invoice number, amounts and tax — pulled into a clean row of data.
Process expense receipts
Read photos or scans of receipts and extract merchant, date, amount and category.
Build a structured log
Turn a folder of documents into a consistent table ready for review or import.
Without Gemini
Open each PDF invoice individually, find the total and invoice number, and type them into a spreadsheet row by row — a slow task that adds up across dozens of documents a week.
With Gemini
Share the invoices with Gemini and ask it to extract the fields into a table with columns like Supplier, Date, Invoice No., Total — ready to review and paste into Sheets.
Invoice extraction with Gemini works best as a short, repeatable process. These steps describe a typical workflow for a batch of supplier invoices or expense receipts.
1
Gather the documents in one place
Collect the invoices or receipts — PDFs, scans or photos — in a single Google Drive folder so Gemini can access them together.
2
Define the fields you need
Decide on the exact column names for your table — for example Supplier, Invoice Date, Invoice Number, Net Amount, VAT, Total, Currency.
3
Ask Gemini to extract into a table
Open the documents with Gemini and ask it to read each one and return the values for your defined fields in a consistent table format.
4
Review against the source documents
Spot-check a sample of extracted rows against the original invoices, paying particular attention to amounts, dates and currency.
5
Paste into Sheets and finalise
Copy the reviewed table into your Google Sheet, applying any number or date formatting before it feeds into reporting or your accounting system.
Note: Gemini reads the document content — it does not connect directly to your accounting software. The output is a table you review and then import, paste or upload using your normal process.
The quality of an extraction depends heavily on how clearly you define the output format. These prompts and checks help keep accuracy high.
Prompt 1 — Extract fields from a batch of invoices
Act as a finance assistant. I am sharing a set of supplier invoices (PDFs or images).
For each invoice, extract the following fields and return them as a table with one row per invoice:
- Supplier name
- Invoice number
- Invoice date (format: YYYY-MM-DD)
- Net amount
- Tax amount
- Total amount
- Currency
If any field is not present on a document, write "Not found" rather than guessing. Flag any document where the total does not appear to equal net plus tax.
Prompt 2 — Categorise expense receipts
Act as a finance assistant. I am sharing a set of expense receipts (photos or scans).
For each receipt, extract:
- Merchant name
- Date
- Amount
- Currency
- Suggested expense category, chosen from this list: [Travel, Meals, Accommodation, Office Supplies, Software, Other]
Return the results as a table with one row per receipt. If the date or amount is unclear or unreadable, write "Check manually" instead of guessing.
Accuracy checklist before using extracted data
Spot-check a sample of rows against the original document images, not just the table
Confirm currency has been identified correctly, especially for international suppliers
Check that dates have not been swapped between day and month formats
Look for any rows marked "Not found" or "Check manually" and resolve them by hand
Confirm totals reconcile with net plus tax before relying on the figures
Important: Treat Gemini's extraction as a fast first draft, not a final record. Always review extracted figures against the source document before they are used in payments, reconciliations or financial reports.