Any document. Any language. Same JSON out. Tell DoXroute what fields matter and drop your files — invoices, customs forms, scanned receipts.
Free tier · 50 documents/month · no credit card
Classification, extraction, splitting, review — the full pipeline runs without writing a single rule.
Write "I need vendor, date, line items and total from my invoices." DoXroute builds the full extraction template.
Create an extractor once from a plain-text description. Reuse it on every future upload without re-describing. Build your own library.
Don't know what the document is? The classifier reads the first page and picks the right schema before extraction starts.
Select up to 50 files, hit upload. A progress bar tracks each one. Download all results when the batch finishes.
Got a 40-page PDF with three invoices and a packing list glued together? DoXroute splits it at the seams and processes each part alone.
Click any extracted value to see where it came from in the original document. Bounding boxes overlay the PDF page.
One API key authenticates everything. Fourteen endpoints cover extraction, batching, splitting, schema management and job polling.
# Extract structured data from a document curl -X POST https://doxroute.com/api/v1/extract \ -H "X-API-Key: YOUR_KEY" \ -F "file=@invoice.pdf" \ -F "doc_type=invoice" # Poll for results curl https://doxroute.com/api/v1/jobs/{job_id} \ -H "X-API-Key: YOUR_KEY"
The same endpoint handles a single invoice or a morning's worth of customs paperwork. Batch mode tracks each file independently.
Send a plain-text description to /extractors, get back a reusable schema ID. Use it on every future upload without re-describing.
Extraction runs on Anthropic Claude. If that's unreachable, the request silently reroutes through OpenRouter. You never see the switch.
HMAC-SHA256 signed webhooks notify on completion. Export results as JSON, CSV, or XLSX per job or in bulk.
| Feature | DoXroute | DocuPipe | Textract | GPT-4o |
|---|---|---|---|---|
| Structured extraction | ✓ | ✓ | ✓ | ✓ |
| Custom schema (natural language) | ✓ | ✓ | — | — |
| Batch upload | ✓ | ✓ | ✓ | — |
| Document splitting | ✓ | ✓ | — | — |
| Visual review + bounding boxes | ✓ | ✓ | — | — |
| Open pricing | ✓ | ✓ | ✓ | ✓ |
TLS everywhere. HSTS headers. No plaintext, ever.
Full data export and account deletion. Self-service, no ticket needed.
Bcrypt-hashed keys, 24h rotating tokens, brute-force lockout after 5 attempts.
30 req/s per client. Content Security Policy headers block injection attacks.
The whole loop takes under two seconds. Try it with the free tier — no credit card, no setup.