tetrel Logo tetrel Logo

docuduct platform

Extracting structured data from complex documents

Self-sufficient data science teams develop turnkey solutions for document analysis with the docuduct platform.

>800 text analysis models available
>100 million documents analyzed
∅ <4 weeks to proof of concept
End-to-end workflow for developing Natural Language Processing applications.
Training with few examples
Meet compliance requirements through redundant processes
Processing of long documents
Prebuilt docuduct modules
Cloud-based or on-premise
DSGVO compliant
Roles, rights and teams
OCR in >100 languages integrated
Connection to common DMS and RPA systems
No NLP engineering & software development necessary

Functionality

The missing link

Accurate data is the raw material for digital transformation and the automation of knowledge-based processes. Our customers analyze documents in seconds and generate structured data for further processing. In doing so, they rely on deep learning based document analysis applications, which can be quickly set up with the docuduct platform and can easily be integrated into existing services via API.

platform architecture

Features

Based on practical experience for more productivity

Building, operating and using document analysis applications from within a single application. The modular design of the docuduct platform has been honed in practice. Our customers use this tool to quickly and directly create document analysis solutions that contribute to their business success. Without engineering and software development.

Upload documents

in-app document conversion

Integrated document processing supports PDF, image documents, Microsoft Office, JSON and text

Multilingual text recognition (OCR)

Scanned documents in >100 languages are converted into machine-readable text by the integrated OCR engine

Train NLP models

Fully automated training

Calibration, windowing, scoring, etc. are automated - our customers can focus on selecting the best training data and models

Integrated Active Learning workflow

NLP models can be trained with up to 90% less labeled examples, for the same model performance

Document Analysis

Classify Documents and Extract Structured Data

Name Entity Recognition (NER) and classification models for long documents can be trained and used in the app

Automated Model Deployment.

Trained NLP models can be deployed as a documented API endpoint with one click

Result control

Integrated “human-in-the-loop” workflow

For redundant & compliance-compliant processes, a human-in-the-loop workflow is automatically integrated for newly deployed docuduct modules

Performance Monitoring

The analysis quality of the NLP models in production is continuously monitored

Export

Format-independent download of results

Download annotated documents in common formats or transfer structured data to downstream systems

Flexible model utilization

Download trained models and use them in other NLP applications

Implementation

Tailor-made for all needs

IT roadmaps and the current situation are taken into account, as are compliance rules. Independent of existing system landscapes and adaptable for any future changes.

Cloud or on-premise
Process your data wherever your business demands it: in the cloud or strictly on-premise.
Flexible integration
Use docuduct modules that have already been developed in existing services via API integration or in common document management and workflow systems.
Scalable
Scale with your requirements: from the first proof of concept to a cluster with several hundred real-time services.