Recipe

Data extraction from receipts

Extract structured line-item data from receipt images using Vision with JSON Schema constrained output. Parse merchant names, totals, tax breakdowns, and individual line items into a typed response.

Visionjson_schemaStructured OutputOCR

How it works

01

Upload receipt

Send a base64-encoded image or pass a URL to the Vision endpoint.

02

Apply schema

The json_schema parameter constrains the model to return only valid Receipt JSON.

03

Parse response

Deserialize the typed response directly into your application models.

Use cases

  • Expense report automation — extract totals and categories from employee receipts
  • Bookkeeping ingestion — parse vendor, date, and line items into accounting software
  • Tax preparation — isolate taxable amounts and tip totals from restaurant receipts