Document AI Konfuzio

Use the document AI Konfuzio Connector to OCR, extract, and retrieve structured information from PDFs, images, handwriting, and scans. Register on https://app.konfuzio.com for free to OCR PDFs and images from various file types, see https://dev.konfuzio.com/web/api.html#supported-file-types. In addition, you can train your custom Document AI following the Tutorial on https://help.konfuzio.com/tutorials/quickstart/ to extract structured information from documents.

This connector is available in the following products and regions:

OCR, extract, validate, process and understand the information in documents by connecting to Konfuzio using this connector. Train Konfuzio to understand any file-based carrier of information to automate complex back-office processes and use more data to generate insights. Konfuzio enables business users who use this connector to categorize documents and extract information. Optionally, data scientists can use Konfuzio as an automated text and image labeling tool with a user-friendly web interface to maintain high-quality data sets to build world-class AI, integrate their AI and improve the AI with humans in the loop.

Prerequisites

1. Register for free on app.konfuzio.com. The free plan provides you with a limited number of features. Have a look at the features of our fee-based service here. If you want to upgrade to a fee-based service please contact us via info@konfuzio.com.

2. Set-up a project as described in our quickstart tutorial on help.konfuzio.com.

How to get credentials

Use the username and password which you used to create a user account on app.konfuzio.com as credentials to use the connector.

Get started with your connector

Find the latest step-by-step process for getting started with your connector on help.konfuzio.com.

Known issues and limitations

• The connector supports ca. 70 languages, see dev.konfuzio.com
• The connector supports various file types. Have a look at dev.konfuzio.com for a detailed list of supported file types.
• The connector supports data normalization for strings that refer to numbers, percentages, date values and booleans, see dev.konfuzio.com

Common errors and remedies

Contact us via support@konfuzio.com in case of errors.

FAQ

We publish FAQ about the Power Automate Connector on help.konfuzio.com.

Creating a connection

The connector supports the following authentication types:

Default

Applicable: All regions

Parameters for creating connection.

This is not shareable connection. If the power app is shared with another user, another user will be prompted to create new connection explicitly.

Throttling Limits

Actions

import requests
import json
from requests.auth import HTTPBasicAuth

url = f"https://app.konfuzio.com/api/v2/docs/{DOC_ID}/"
auth = HTTPBasicAuth(KONFUZIO_USER, KONFUZIO_PASSWORD)
data = requests.get(url=url, auth=auth).json()

# Mark Extraction as correct.
data['labels']['Bruttozahlweise']['extractions'][0]['correct'] = True

# Add new Extraction which has not been in result list.
data['labels']['Bruttozahlweise']['extractions'].append({'value': '123,45'})

r = requests.patch(url = url, data=json.dumps(data), auth=auth, headers={'Content-Type': 'application/json'},)

Delete a document

Give feedback to the extraction results of a document

import requests
import json
from requests.auth import HTTPBasicAuth

url = f"https://app.konfuzio.com/api/v2/docs/{DOC_ID}/"
auth = HTTPBasicAuth(KONFUZIO_USER, KONFUZIO_PASSWORD)
data = requests.get(url=url, auth=auth).json()

# Mark Extraction as correct.
data['labels']['Bruttozahlweise']['extractions'][0]['correct'] = True

# Add new Extraction which has not been in result list.
data['labels']['Bruttozahlweise']['extractions'].append({'value': '123,45'})

r = requests.patch(url = url, data=json.dumps(data), auth=auth, headers={'Content-Type': 'application/json'},)

Retrieve the extraction results for a document

Upload a new document