Turn any document into structured data for your ERP
UnchainDoc extracts, classifies, and delivers data from invoices, contracts, bank statements, and any other document — directly into Odoo, SAP, NetSuite, and more. Via REST API in JSON.
From inbox to ERP record in four steps
One pipeline, no templates to maintain. Drop documents in any way you like — get validated, typed data out.
Ingest
Forward an email, push to the API, or sync a Drive / SharePoint folder. PDFs, scans, photos, ZIPs.
Classify
Auto-detect the document type and split multi-document batches into separate files.
Extract
Header fields, nested tables, line items, stamps and handwriting — each with a confidence score.
Deliver
JSON over REST or webhook, mapped straight onto the objects in your ERP. No middleware scripts.
Built for documents that break other tools
Crumpled scans, 40-page contracts, handwritten delivery notes — all handled by the same endpoint.
99%+ extraction accuracy
Field-level confidence scores let you auto-post what is certain and route the rest for review.
Handwriting recognition
Signatures, hand-filled forms and margin notes read in 30+ languages, print and cursive.
Auto-classification & splitting
One 90-page scan becomes 12 typed documents, each routed to the right pipeline automatically.
Table & line-item extraction
Multi-page tables keep their structure — rows, spans and totals arrive as nested arrays.
Human-in-the-loop review
A review queue with side-by-side source and fields. Every correction retrains your model.
Custom AI models in minutes
Label five samples of your own document type and ship a dedicated model the same day.
Plugs into your stack
Native connectors for the systems your team already runs on — or one REST endpoint if you prefer to wire it yourself.
{
"document_id": "doc_8f24ac",
"doc_type": "invoice",
"invoice_number": "INV-2026-04417",
"date": "2026-04-17",
"vendor": {
"name": "Northwind Supply GmbH",
"vat_id": "DE811907980"
},
"currency": "EUR",
"total": 4820.50,
"line_items": [
{
"description": "Aluminium bracket 40x40",
"qty": 120,
"unit_price": 18.40,
"amount": 2208.00
},
{
"description": "Freight, DAP Rotterdam",
"qty": 1,
"unit_price": 312.50,
"amount": 312.50
}
],
"confidence": 0.994,
"review_required": false
}
Where teams put it to work
Finance & Accounting
Close the month without typing. AP invoices matched to POs, statements reconciled, expenses coded before they hit the ledger.
Logistics & Trade
Shipment paperwork keyed the moment it lands — including scans from the cab and hand-signed delivery notes.
Legal & Compliance
Clause-level contract data, obligation dates and KYC identity checks extracted into a reviewable, auditable record.
“We went from three people keying invoices to one person approving them. The confidence scores are what made us trust auto-posting.”
Head of Finance Ops, Kolbruk
“Bills of lading arrive as phone photos at 2am. UnchainDoc reads them anyway and the data is in our TMS before the driver is home.”
Operations Director, Atlas Freight
“Two engineers, one afternoon, and a custom model for our KYC packets. The API was the easiest part of the whole project.”
Compliance Lead, LexForm
Start free, scale by pages
No per-seat pricing. No template setup fees.
Frequently asked
How is my data secured, and are you GDPR compliant?
Documents are encrypted in transit and at rest, processed in your chosen region (EU or US), and never used to train shared models. We are ISO 27001 certified, GDPR compliant with a standard DPA, and support configurable retention — including delete-on-extract for zero storage.
Which file formats do you support?
PDF (native and scanned), JPG, PNG, HEIC, TIFF (multi-page), DOCX, XLSX, EML with attachments, and ZIP archives. Photos taken on a phone are fine — skew, shadows and low contrast are corrected before extraction.
What are the API rate limits?
Free plans allow 60 requests/minute; Pro allows 600 requests/minute with burst queueing and async batch endpoints for large uploads. Enterprise limits are set per contract. Every response includes X-RateLimit-Remaining headers.
Do you really read handwriting?
Yes — printed and cursive handwriting in 30+ languages, including hand-filled form fields, quantities on delivery notes and signature detection. Handwritten values return a lower confidence score by default so they can be routed to human review automatically.
How are custom models trained?
Upload around five examples of your document type, click the fields you want, and training completes in minutes — no data science team required. Corrections made in the review queue feed back into the model, so accuracy improves as your team works.
Ready to unchain your documents?
Process your first 50 pages today. Connect your ERP this week.