One API for documents and the web
Invoices, receipts, forms, IDs — and any live URL. Point a document at
POST /api/v1/extract, a page at
/api/v1/screenshots, or do both: capture
a URL, then extract its data in one call.
New — screenshot any URL, then extract its data
Turn any document or website into structured data
dOCR captures PDFs, images, and live web pages — and returns the exact fields you define as JSON. Skip the browser cluster and the OCR-plus-parsing pipeline; call one API.
$ curl -X POST docr.dev/api/v1/screenshots -d '{"url":"stripe.com"}'
invoice.pdf PDF
stripe.com
output.json from invoice.pdf
{
"title": "INV-2042",
"amount": "1,280.00",
"date": "2026-07-15"
}
structured JSON, ready to use
Trusted by teams processing millions of documents
Capabilities
Everything you need to capture and extract
dOCR owns the messy parts — rendering pages and parsing documents — so you get clean screenshots and typed, validated JSON instead of a browser cluster and a wall of OCR text.
Structured JSON, not OCR text
dOCR returns the typed fields you defined — each one located, validated, and scored — so you write zero parsing code.
invoiceNumber "INV-2042" 99%
total 1280.00 98%
dueDate 2026-07-15 96%
Screenshot any URL
Render any page to PNG, JPEG, WebP, or PDF — full-page, any device viewport, retina. dOCR blocks ads and cookie banners for you. No headless browser to babysit.
stripe.com captured · png example.com captured · pdf news.ycombinator.com rendering…
Fast on every page
Most documents come back in a couple of seconds — multi-page PDFs included — so dOCR fits inside a request, not a nightly batch job.
Document types you control
Define the fields you want once as a
documentType, and every extraction
returns that exact shape — same keys, same types, every time.
Dashboard or API
Upload in the dashboard to try things out and browse Extraction History, then automate
the same extraction from your backend — one
POST, no infrastructure to run.
Developer experience
One request — a screenshot or structured data
No SDK to learn, no browser cluster, no model to host. Call a single endpoint from any language and dOCR returns a rendered page or the fields you defined as JSON.
- Extract documents or screenshot any URL — one key, one endpoint family
- Capture a page, then pull structured data from it in a single call
- Typed JSON with a confidence score on every field — no parsing code
const form = new FormData();
form.set('file', await openAsBlob('invoice.pdf')); // PDF, image, or DOCX
form.set('documentType', 'invoice');
const res = await fetch('https://docr.dev/api/v1/extract', {
method: 'POST',
headers: { Authorization: `Bearer ${process.env.DOCR_API_KEY}` },
body: form,
});
const { data } = await res.json();
// → { invoiceNumber: 'INV-2042', total: '1280.00', dueDate: '2026-07-15' }const res = await fetch('https://docr.dev/api/v1/screenshots', {
method: 'POST',
headers: {
Authorization: `Bearer ${process.env.DOCR_API_KEY}`,
'Content-Type': 'application/json',
},
body: JSON.stringify({ url: 'https://stripe.com', fullPage: true, blockCookieBanners: true }),
});
const { screenshot } = await res.json();
// → { id: 'cap_…', status: 'ready', format: 'png', url: 'https://…' }curl https://docr.dev/api/v1/extract \
-H "Authorization: Bearer $DOCR_API_KEY" \
-F file=@invoice.pdf \
-F documentType=invoiceTransparency
See exactly how each field was found
Every extraction is traced from upload through OCR, field extraction, and schema validation — with structured logs alongside. When a field looks off, you know which stage produced it and how confident dOCR was.
extraction 01J9JD4QK7 POST /api/v1/extract 200 1.4s invoice
Trace waterfall
POST /api/v1/extract 1.4s upload.validate 15ms ocr.pages 0.42s extract.fields 0.73s validate.schema 60ms serialize 12ms
logs — streaming
- 12:04:31.118 INFO request.start documentType=invoice
- 12:04:31.140 INFO upload.ok pages=2 bytes=184kb
- 12:04:31.620 INFO ocr.ok pages=2
- 12:04:31.980 WARN extract.lowconf field=dueDate score=0.71
- 12:04:32.360 INFO extract.ok fields=7
- 12:04:32.520 INFO request.end status=200 dur=1.40s
Every extraction is logged to your dashboard — confidence scores, page counts, and timings, no setup required.
Customers
Teams that switched, stayed
“We deleted an entire OCR-and-regex pipeline. dOCR returns the exact fields we defined as clean JSON — invoice processing that took our team days now runs in seconds.”
“Custom document types were the moment it clicked. We described the fields once, and accuracy on our messy scanned receipts jumped immediately.”
“We prototyped in the dashboard in an afternoon, then moved the same extraction to the API in a few lines. No model to train, no infrastructure to run.”
Capture and extract, from one API key
Screenshot a URL or pull structured data from any document — same key, same endpoint family. Start free — no credit card, no pipeline to build. Pay as you scale.