DOCUMENT AI · FOR EVERY TEAM

Turn invoices, IDs and forms into structured JSON.

A document extraction API that reads key-value fields and full tables, then returns clean JSON with named fields and bounding boxes. Efficient by design, fewer server cycles and less data moved per page, so accurate extraction stays affordable for startups, integrators and enterprises alike.

Try now API reference →
Key-value + table extraction Clean JSON with bounding boxes Priced for every team
SUPPORTED DOCUMENT TYPES

Pre-trained for the documents you process every day

Invoices

Bank statements

Receipts

KYC forms

Contracts

Payslips

Lab reports

ID documents

Don't see your document type? Document Analysis API supports custom rule-based extraction. Read more →

EXAMPLE RESPONSE

Clean JSON, ready for your ERP / DB

python
result = client.extraction.basic(
    file="bank_statement.pdf",
    document_type="bank_statement"
)

print(result.fields)
response.json
{
  "account_holder": "Rakesh Sharma",
  "account_number": "XXXX1234",
  "bank": "HDFC Bank",
  "ifsc": "HDFC0000123",
  "period": {
    "from": "2026-04-01",
    "to": "2026-05-31"
  },
  "opening_balance": 45000.00,
  "closing_balance": 82340.50,
  "transactions": [...112 records]
}
USE CASES

Where document extraction ships

AP automation

Vendor invoice → JSON → ERP. Cuts manual data entry. Pairs with Tally / Zoho / SAP.

Lending underwriting

Bank statement extraction → 60+ data points for credit scoring. Sub-30-sec turnaround.

Expense management

Mobile-captured receipt → structured expense entry. T&E automation made simple.

CKYC onboarding

Pre-fill customer onboarding from KYC document. CKYC 2.0-ready.

Contract intake

Pull party names, dates, payment terms from contracts before legal review.

Insurance claims

Hospital bill structure → claim line items. Speed up TPA processing.

PRICING

Flat per-page rate

Per page processed. Multi-page documents counted per page. Same flat rate regardless of document type or field count. Pricing shown in your local currency, change country in the top-right to switch.

See full pricing →

Affordable to all

The flat per-page rate is low because the extraction pipeline is optimized for time and space first, fewer server cycles and less data moved and stored per page. The cost savings follow the efficiency, with no drop in output quality, so any team can run production extraction. See the full Document AI API range →

Extract Pro tier

The Pro tier on this API adds confidence scores, validation rules, and guardrails, for fintech and regulated workflows where false positives are costly. Same JSON contract, more assurance per field. Talk to us about Pro →

FAQ

Common questions

What does the document extraction API do?
It reads documents and returns their data as structured JSON. The API pulls key-value fields and full tables from documents such as invoices, IDs, and forms, and returns clean JSON with named fields and bounding boxes, so the output drops straight into an ERP or database without manual data entry.
How do I extract data from an invoice or PDF into JSON?
You send the file along with its document type, and the API returns the extracted fields as JSON. For a bank statement, for example, it returns named fields such as account holder, account number, opening and closing balances, and the full list of transactions. Multi-page PDFs are processed and counted per page.
Which document types can it extract out of the box?
The API is pre-trained for invoices, bank statements, receipts, KYC forms, contracts, payslips, lab reports, and ID documents. You choose the document type and it returns the relevant fields for that document. For types outside this set, the Document Analysis API supports custom rule-based extraction.
Does it extract tables and line items, or only single fields?
It extracts both key-value fields and full tables. Multi-row data such as invoice line items or bank-statement transactions is returned as an array of records in the same JSON response, alongside the top-level named fields.
What format is the extracted data returned in?
Clean JSON with named fields and bounding boxes. Each extracted value comes back under a named field, and fields carry the bounding box of where they were found on the page, so results are ready for an ERP or database and easy to verify against the source document.
How is the document extraction API priced?
On a flat per-page rate. Every page is billed at the same flat rate regardless of document type or the number of fields extracted, and multi-page documents are counted per page. Prices are shown in your local currency; see the Document AI pricing page for details.
What does the Extract Pro tier add?
The Pro tier adds confidence scores, validation rules, and guardrails on top of the same JSON contract. It targets fintech and regulated workflows where false positives are costly, giving more assurance per field without changing the field structure. Talk to us about Pro for access.
How can I test the document extraction API before integrating?
You can try it with a free trial that includes signup credit, no integration required. Upload a sample document and see the real JSON output, then use the API reference to wire it into your workflow.
THE FULL TOOLKIT

Explore the rest of the Document AI API

One license, every capability — take this and you also get all of these.

Data Extraction OCR Data Masking Document Analysis

Test document extraction with the free trial

Free signup credit. Upload a sample, see real JSON output.

Try now All Document AI APIs