Finance & Admin
Expense & Receipt Auto-Logger
Forward a receipt email or snap a photo — it gets read, categorized, and logged.
Saves about ~2 hrs per week
What it does
- Reads receipts and invoices from email or a photo folder
- Extracts amount, vendor, and date with AI vision
- Categorizes the expense into your categories
- Appends rows to a sheet or accounting tool
How it works
- Receipt email or photo lands in the watch zone
- AI extracts the amounts and vendor
- Expense categorized automatically
- Row appended to the ledger
What's included
- Receipt capture
- AI extraction
- Ledger connection
Set it up yourself
Follow these 4 steps in order. You will need accounts for: Gmail, Sheets.
Step 1: Create the watch inbox or folder
- In Gmail create a 'Receipts' label and a rule that applies it to anything you forward (just forward the email to yourself with it — or better, create a filter for common senders).
- Alternatively create a Drive folder 'Receipt Photos' for scans.
- Make sure the workflow watches both spots.
- Optional: give the label a rule so vendor receipt emails auto-tag.
- Test that an email you forward lands under the label.
Watch out: Use one label + forwarding as the reliable path — relying on sender auto-detection misses half the vendors.
Step 2: Add the extraction prompt
- Build the prompt (vision-capable model): 'Read this receipt. Return JSON: {vendor, amount, currency, date, category, items[]}. Amount as the final total.'
- From email: pass the attachment to the image model.
- From photo: pass the image straight through.
- Add a fallback category 'uncategorized' when classification is unsure.
- Paste with API key and test on 3 receipts of different layout.
Watch out: Images of receipts vary wildly in layout — test on crumpled, store-printed, and digital receipts; the vision model must handle all three or the log will silently miss.
Step 3: Map your categories
- List your expense categories (Meals, Travel, Software, Office, etc.).
- Paste them into the prompt so the AI classifies against them.
- Add vendor→category overrides (e.g. 'Stripe' → Software) in a sheet for accuracy.
- Test the category mapping on 10 real receipts.
- Adjust overrides where the AI guesses wrong.
Watch out: The vendor-override sheet is what makes this near-perfect — the AI learns your regular vendors through rules faster than through prompt tuning.
Step 4: Connect the ledger
- Create the 'Expense Ledger' sheet with the exact columns you need (Date, Vendor, Amount, Category, Source, Note).
- Have the workflow append a row per receipt.
- For accounting tools (QuickBooks/Xero), use their API to create the expense instead.
- Test the append with a real receipt.
- Weekly: review the ledger for gaps.
Watch out: Match the ledger columns to your accountant's import format — a clean export that imports straight into your books is the delivery goal.
Frequently asked questions
What does the Expense & Receipt Auto-Logger automation do?
Forward a receipt email or snap a photo — it gets read, categorized, and logged. Reads receipts and invoices from email or a photo folder
How much time does Expense & Receipt Auto-Logger save?
It is estimated to save ~2 hrs per week. Actual savings depend on your volume.
Which tools does Expense & Receipt Auto-Logger use?
It works with Gmail and Sheets.
What does Expense & Receipt Auto-Logger include?
Receipt capture, AI extraction, Ledger connection.
Can I set up Expense & Receipt Auto-Logger myself?
Yes. The do-it-yourself guide on this page walks through 4 steps, from getting access to testing the finished automation.