Use case / Expense operations

Receipt fields without manual rekeying.

Define the small, stable receipt object your expense workflow needs instead of wiring downstream systems to raw OCR text.

Start free See a real request and response

Example schema

Ask for the fields that move work forward.

Field names, types, and descriptions travel with every extraction request, so the response matches the object your code already expects.

Schema fields5 fields

merchant_nametransaction_numbertransaction_datetotal_amountpayment_method

Response

What your integration receives.

The terminal result for the schema above. Validate it against your own types before writing to a system of record.

extract.tsTypeScript client

const { result } = await client.extractFile({
  file,
  idempotencyKey,
  input: { fileName, mediaType, sizeBytes },
  schema: {
    version: "1",
    fields: [
      { name: "merchant_name", type: "string" },
      { name: "transaction_number", type: "string" },
      { name: "transaction_date", type: "string" },
      { name: "total_amount", type: "number" },
      { name: "payment_method", type: "string" }
    ]
  }
});

terminal resultcompleted

{
  "merchant_name": "Northstar Cafe",
  "transaction_number": "TXN-7314",
  "transaction_date": "2026-07-02",
  "total_amount": 18.75,
  "payment_method": "Visa"
}

Production path

From document to validated workflow input.

Choose the fields needed for reconciliation.
Upload a supported receipt image with tenant authentication.
Submit an asynchronous job using the same schema contract.
Route missing or uncertain values to review.

FAQ

Frequently asked questions

Can I choose which receipt fields are returned?

Yes. The submitted schema defines the expected field names, types, descriptions, and requirements.

Does Routeser replace expense-policy validation?

No. Extraction produces structured data; your application remains responsible for expense policy and approval decisions.

How are receipt totals typed?

Declare monetary fields as numbers. An explicit type is what makes downstream normalization deterministic.

Next step

Run this schema on your own documents

Open the console to define these fields in a workspace and evaluate them against real documents.

Open console