Document Intelligence

Overview

Document Intelligence accurately classifies any PDF document and extracts key information — all with exclusively in-memory processing and a zero-retention architecture that guarantees your sensitive data never leaves Switzerland.

Security and Data Privacy

GuardOS is committed to ensuring the security and privacy of your data. See Security and Data Privacy for more information.

How It Works

Document Intelligence Schema

Document Intelligence applies a multi-step, mixture of experts approach to extract key information from a given PDF document.

Step 1 - Pre-Processing

Each document undergoes an initial pre-processing stage to remove noise and optimize it for accurate classification.

This step includes textbox- and image-extraction. The textbox-position and size, textsize and the text itself are extracted. Any images inside the PDF are photometrically cleaned, geometrically normalized and denoised. The contrast and illumination are adapted. Then the image gets grayscaled, resized and morphological operations are applied.

Step 2 - Data Extraction

To deliver the most accurate results, each document is intelligently processed using a cutting-edge mixture-of-experts approach. Specialized models, ranging from machine learning and natural language processing to OCR and large language models, are strategically applied to specific tasks for maximum effectiveness.

For instance, determining whether a document includes a signature is handled by a custom-trained computer vision model designed for high precision.

Step 3 - Post-Processing

In the post-processing phase, extracted information is verified, refined, and consolidated into a clean, standardized format.

This step ensures consistency across outputs by validating key data points, resolving ambiguities, and applying business logic as needed.

The result is structured, reliable information—ready for seamless integration into downstream systems or workflows.

Input File Formats

Document Intelligence supports the following input document file types:

• PDF (up to 15MB, unlimited pages)

Pricing

Pricing is based on expected volume. Please contact us for a quote.

Supported Document Types

The Document Intelligence API currently supports the following document types:

Preview Stage Document Types

The following document types are currently in preview.

Supported Document Metadata

The Document Intelligence API currently supports the following document type agnostic metadata:

Endpoints

POST: /api/v1/document-intelligence

Request Headers

interface DocumentIntelligenceRequest {
	'Content-Type': 'multipart/form-data'
	'x-api-key': string
}

Request Body

The request should be a multipart/form-data with a single file field containing a PDF document.

Success Response

interface DocumentIntelligenceResponse {
	document: {
		type: 'commercial_register' | 'rental_agreement' | 'loan_agreement' | ... | 'other'
		metadata: {
			companyName: string | null
			companyRegistrationNumber: string | null
			documentDate: string | null // Format: YYYY-MM-DD
			hasSignature: boolean
		}
		pages: number // Number of pages in the document
	}
}

Error Response

interface DocumentIntelligenceError {
	error: string
	details?: string
}

Examples

Example Request

curl -X POST https://api.guardos.ai/api/v1/document-intelligence \
  -H "x-api-key: YOUR_API_KEY" \
  -F "file=@document.pdf"

// Using a file input element: <input type="file" id="pdfFile" accept="application/pdf">
const fileInput = document.getElementById('pdfFile');
const file = fileInput.files[0];

const form = new FormData();
form.append('file', file);

const response = await fetch('https://api.guardos.ai/api/v1/document-intelligence', {
  method: 'POST',
  headers: {
    'x-api-key': 'YOUR_API_KEY'
  },
  body: form
});

const result = await response.json();
console.log(result);

import fetch from 'node-fetch';
import { createReadStream } from 'fs';
import FormData from 'form-data';

async function analyzeDocument() {
  const form = new FormData();
  form.append('file', createReadStream('document.pdf'));
  
  const response = await fetch('https://api.guardos.ai/api/v1/document-intelligence', {
    method: 'POST',
    headers: {
      'x-api-key': 'YOUR_API_KEY',
      ...form.getHeaders()
    },
    body: form
  });
  
  const result = await response.json();
  console.log(result);
}

analyzeDocument();

import requests

url = 'https://api.guardos.ai/api/v1/document-intelligence'
headers = {
    'x-api-key': 'YOUR_API_KEY'
}
files = {
    'file': ('document.pdf', open('document.pdf', 'rb'), 'application/pdf')
}

response = requests.post(url, headers=headers, files=files)
print(response.json())

Example Response

{
	"document": {
		"type": "commercial_register",
		"metadata": {
			"companyName": "Acme Corporation AG",
			"companyRegistrationNumber": "CHE-123.456.789",
			"documentDate": "2023-05-15",
			"hasSignature": true
		}
	},
	"pages": 3
}