πŸ“„
Available

Doc Extract

Doc Extract combines a high-fidelity OCR pipeline with specialized language models to automatically extract key information from your documents. The service handles invoices (line items, amounts, VAT), resumes (skills, experience, education), ID documents (name, date of birth, number), and certificates (dates, signatories, references). Extracted data is returned as structured JSON, ready to integrate into your information system.

OCRLLMInvoicesResumes

How it works

1

Document upload

Send your PDF, image, or DOCX file via the REST API. The service automatically detects the document type.

2

OCR and pre-processing

The document is analyzed by our OCR pipeline, which corrects orientation, improves contrast, and extracts raw text.

3

LLM extraction

A specialized language model analyzes the extracted text and identifies the relevant fields based on the detected document type.

4

Structured data

Extracted data is returned as structured JSON with a confidence score per field, ready for integration.

API call example

extract_invoice.py
import httpx

response = httpx.post(
    class="text-teal">"https:class="text-text-faint italicclass="text-teal">">//api.djinn.dev/v1/extract",
    headers={class="text-teal">"X-API-Key": class="text-teal">"djinn_live_..."},
    files={class="text-teal">"file": open(class="text-teal">"invoice.pdf", class="text-teal">"rb")},
    data={class="text-teal">"document_type": class="text-teal">"invoice"},
)

result = response.json()
print(fclass="text-teal">"Vendor: {result['vendor_name']}")
print(fclass="text-teal">"Total (incl. tax): {result['total_ttc']} EUR")
print(fclass="text-teal">"Confidence: {result['confidence']:.0%}")

for line in result[class="text-teal">"lines"]:
    print(fclass="text-teal">"  - {line['description']}: {line['amount']} EUR")

Technical specifications

FormatsPDF, JPEG, PNG, TIFF, DOCX
Max size50 MB
P95 latency< 3s
SLA99.9%

Ready to integrate this service?

Test the API free with 1,000 calls included. Integrate in minutes.