Document AI

Google Document AI Alternative from Germany

Konfuzio - Starter Kit
All information about the API, prices, contracts and a short presentation can be emailed to you within seconds.

Train even complex documents via the web browser

We also define annotating documents as teaching the AI. From about seven annotated documents, the AI learns. How you Annotations correctly and thus quickly teach the AI what to do with just a few document examples, you will learn in the following three tutorials.

  • Define categories of documents yourself
  • Customize field names
  • Read individual information and paragraphs
  • Read tables
  • Recognize titles, paragraphs, images, tables and lists without training
  • Choose from different OCR engines
  • Start immediately after free registration

Link Document AI(s) to existing systems as a REST API

Use Konfuzio's open infrastructure to connect your systems directly via interface. Run your Document AI(s) for multiple API endpoints, securely separating data by users, groups and projects.

  • Bounding boxes of letters, lines and more
  • Bounding box & character offset of the annotation
  • Features enrichment by converting a character offset from NLP into a bounding box
  • Applying new annotations via API
  • Feedback API for Human in the Loop
  • Versioned annotation incl. creator, controller & etc.
  • Upload of new documents & management of originals
  • OCR text incl. HOCR
  • Projects, labels & template metadata
  • Automatic duplicate check for training and test data
  • Generate thumbnails per page
API interface of Konfuzio
Watch Konfuzio API video on YouTube now!

Program your own Document AI with the Konfuzio Python SDK

Own models and access to training data

  • Python API Wrapper
  • Change tokenizer
  • Customize OCR Pipeline
  • Define classification of texts yourself
  • Define Entity Recognition yourself
  • Define semantic parsing yourself
Document AI Python SDK
Arrow-up