Overview

AI receipt scanning lets you upload a photo or image of a receipt and have Valzotra automatically extract the vendor, amount, date, and category. The extracted data pre-fills the expense form so you can review it, make corrections if needed, and save. This feature uses Claude Haiku (by Anthropic) for image analysis and is available on the Host and Portfolio plans.

How it works

  1. Go to Financials > Expenses and click Add expense.
  2. On the upload screen, provide your receipt using any of these methods:
    • Drag and drop an image file onto the upload area.
    • Click Choose file to select an image from your device.
    • Click Take photo to capture a receipt with your device camera.
    • Paste from clipboard using Ctrl+V (Windows) or Cmd+V (Mac).
  3. Valzotra compresses the image (if needed) and sends it to the AI extraction service.
  4. The AI reads the receipt and attempts to extract: vendor name, total amount, currency, date, expense category, and a brief description.
  5. Fields that the AI extracted with sufficient confidence are pre-filled in the expense form. A green banner confirms the pre-fill, or a yellow warning appears if the AI could not read all fields.
  6. Review every pre-filled field, correct anything that looks wrong, and fill in any fields the AI missed.
  7. Click Add expense to save. The receipt image is automatically attached to the expense record.

If you prefer not to use receipt scanning, click Skip — enter manually to go directly to the manual entry form.

Limits & requirements

  • AI receipt scanning requires the Host ($34/month) or Portfolio ($49/month) plan. It is not available on the Comply plan.
  • Supported image formats: JPEG, PNG, and WebP. PDF files are accepted for upload and attachment but cannot be read by the AI — you will need to fill in the fields manually for PDFs.
  • Maximum file size: 10 MB per receipt image.
  • Images are compressed client-side before upload to reduce file size and speed up processing.
  • The AI uses a confidence threshold when pre-filling fields. If a field's confidence is below the threshold, it is left blank for you to fill in manually.
  • One AI extraction call is made per receipt upload. The cost is tracked internally but is included in your Host or Portfolio plan — there is no separate per-scan charge.

FAQ

How accurate is the AI extraction? Accuracy depends on the quality and clarity of the receipt image. Clear, well-lit photos of standard printed receipts typically produce good results. Handwritten receipts, faded text, or unusual layouts may yield partial or inaccurate extractions. Always review the pre-filled data before saving.

What happens if the AI cannot read my receipt? You see a yellow warning banner, and the form fields are left blank. You can fill in the details manually. The receipt image is still attached to the expense so you have it on file.

Can I use receipt scanning on my phone? Yes. The upload screen includes a Take photo button that opens your device camera. You can also paste a screenshot from your clipboard.

Does the receipt image get stored? Yes. When you save the expense, the receipt image is automatically stored as an attachment. You can view, download, or delete it later from the expense detail page.

Why can't the AI read PDF receipts? The AI model used for extraction (Claude Haiku) processes image data. PDF receipt extraction is not yet supported. You can still upload a PDF as an attachment, but the form fields will not be pre-filled.