Capturing Receipts and Invoices with AI

What AI document capture does

StretchBooks includes an AI document reader that turns a photo or scan of a financial document — a receipt, invoice, or bill — into structured data. Instead of typing everything by hand, you upload the image and StretchBooks reads it and hands back the fields, ready to review and save.

What it extracts

From a document image, the AI attempts to read:

• Document type — invoice, receipt, bill, statement, or other.
• Vendor name.
• Invoice/reference number.
• Date.
• Currency.
• Subtotal, tax, and total.
• Line items — description, quantity, unit price, and amount for each.

Amounts are returned exactly as read, and any field the document doesn't show comes back blank so nothing is invented.

Step-by-step: capture a document

1. Open the AI capture action (available where you record expenses and bills).
2. Provide the document image — upload a photo or scan of the receipt, invoice, or bill.
3. StretchBooks reads it and returns the extracted fields.
4. Review the results — check the vendor, date, tax, and total against the original.
5. Use the extracted values to pre-fill an expense or bill, adjust anything the reader missed, choose the account/tax code, and Save.

Worked example

Northlight Coaching gets a paper receipt from the print supplier:

1. Snap a photo of the receipt and run AI capture.
2. StretchBooks returns: vendor Maple Ridge Print Co., date, currency CAD, subtotal $200.00, tax $10.00, total $210.00, and a line for printed workbooks.
3. Review, then create a bill or expense from those values — pick the expense account (or vendor) and the GST tax code, confirm the total matches, and save.

Availability

AI document capture is an optional capability that your operator turns on. When it's not enabled (or the model isn't ready), the feature responds that it's unavailable and you simply enter the document manually — nothing else changes. When it is enabled, it's designed to fail safe: if it can't read an image, it tells you rather than guessing.

Tips

• Photograph flat, well-lit documents. A straight, glare-free image reads far more accurately than a crumpled or shadowed one.
• Always review before saving. The AI extracts; you confirm. Check the total and tax against the paper.
• Let it do the typing, you do the coding. The reader fills amounts and vendor; you still choose the right expense account and tax code.
• Keep the original image as the record's receipt so you have the source at tax time.

Troubleshooting

The feature says it's unavailable. AI capture isn't enabled for your organization, or the model isn't ready. Enter the document manually and ask your operator to enable it.

Nothing came back from an image. The reader couldn't extract anything — retake the photo with better lighting and framing, or enter the details by hand.

Some fields are blank. The document didn't clearly show them. Blanks are intentional (the reader won't guess); fill them in from the original.