Artificial intelligence business solutions for more efficient and automated work processes
OCR WITH HANDWRITING
ANONYMIZATION
ENTITY RECOGNITION
DOCUMENT CLASSIFICATION
ACCESSIBILITY WITH AI
BETTER READING WITH AI
FONT RECOGNITION
OCR WITH HANDWRITING
ANONYMIZATION
ENTITY RECOGNITION
DOCUMENT CLASSIFICATION
CIB deepER: The powerful AI OCR specialising in business documents masters handwriting on scans
CIB deepER is an artificial intelligence based optical character recognition engine that recognizes and extracts text from images and scanned documents.
It is one of the most powerful OCR engines and supports English, German, Spanish and Portuguese.
The OCR can be individually trained and integrated for integration and use in dark processing.
CIB PoP: Data protection for your documents with anonymization and pseudonymization
With CIB PoP we protect sensitive data in documents in accordance with the GDPR and make them retrievable or not, depending on the requirements. Anonymized documents can be used as a basis for further AI training, or published or shared in a way that complies with legal requirements.
The project developed in collaboration with the Fraunhofer Institute, lays the foundation for AI training.
CIB recogNice: Automatic document classification with AI speeding up processes
Our AI-based classification module determines the appropriate classes from the extracted text structure of a document:
- Integrate AI-based decisions into processes via REST API
- Flexibly adjust the level of process automation
- Manually check cases reported as unsafe by the AI
- Feed corrected decisions back into the AI data
CIB sherlock: AI recognition for automated data extraction from documents
CIB sherlock accurately extracts the data relevant to a use case and enables reliable and automated processing with artificial intelligence.
- Specialized for business documents
- AI recognizes whether a document can be automated with dark processing or whether it needs to be controlled manually.
- Self-learning and scalability
Font recognition on scanned documents with the help of artificial intelligence
Our artificial intelligence model for scanned document font recognition detects font, size and style at the word level.
- Detects font, size and color
- Recognizes all common font families
- Powerful recognition at word level
- Individually trainable AI model
CIB superResolution: restores illegible text
Restores text on images, making scans, faxes or photos of documents legible
Our document trained AI examines image material and automatically restores illegible text.
CIB autoCrop: Automatically crop document scans
The AI model CIB autoCrop recognizes the page and automatically trims the edges of scanned documents.
- Automatic page corner detection
- Automatic cropping to the document page
- Automatic perspective correction
- Manual correction if necessary - and the AI learns!
- Business Integration
CIB easyRead: Comfortable reading with AI on small screens, scans and documents
- Easy reading without zooming in
- Also works with photographed or scanned documents
- AI-based text recognition and system
- Available in our app CIB doXicloud for free!
CIB ridmi: Accessible reading and understanding
CIB ridmi will enable barrier-free reading of scanned or photographed files.
Together with the app CIB doXisafe and the AI model CIB deepER texts are extracted from images and prepared barrier-free for screen readers.
CIB autorotate: Automatically correct misaligned documents with AI
CIB autorotate is specially trained for text and business documents, corrects incorrectly scanned document pages, and automatically aligns them.
- Align documents during upload
- Detect errors even in stacks of pages
- Identification of documents with scanning errors
- Automation of workflows in input management
Protection of hypersensitive data
The HILANO project, which is funded by KMU Innovativ, ensures the manual anonymization of documents from the medical and legal sectors that are critical in terms of data protection.
A continuously learning AI recognizes sensitive data and makes suggestions to the human user. The user’s correction of these suggestions trains the AI on-premise, on what the company is looking for in anonymization. The data remains in the hands of the company at all times, which ensures data protection.
In cooperation with: GLANOS, Bucerius Law School, University of Hamburg
CIB smartER: RAG - Mit KI in die Zukunft der intelligenten Dokumentenrecherche
CIB smartER is an innovative research and analysis tool (RAG) that combines state-of-the-art semantic search technologies with powerful Large Language Models (LLMs).
- AI-assisted, synchronized, intelligent, full-text search
- Search for similar documents and text fragments
- With references
- Unlocking valuable content in the archive
- Communicate in natural language
Case studies
CIB deepER makes big AIs even more efficient!
You've Got Mail! Automation in the incoming mail process
Bringing inaccessible databases to life with AI and automation
Migration and automation in dark processing for Beihilfe-Service GmbH
Automation with AI in the application processing of the Regional Administration of Bavaria
Modern technologies, intelligent solutions
Our AI team has extensive knowledge of many different approaches to building neural networks using different frameworks.
Neuronal Networks:
- Long-Short Term Memory
- Convolutional Neural Network
- Generative Adversarial Network
- Multilayer Perceptron
- Transformers
- ResNet
Frameworks:
- Tensorflow
- Keras
- PyTorch
Current CIB AI research areas
We are researching these topics now and in the future with partners or as part of funding programs.
Natural Language Processing
Our AI solutions, we support the creation and understanding of documents wherever possible.
- Language assessment
- Determine language level
- Automatic transcription of video conferences
- Recognition of reading direction and sequence
- Linguistic analysis
- Semantic search
- Automatic text suggestions and alternative formulations
- Automatic conversion of non-structured documents into Accessible PDF/UA
Image manipulation
From scans to smartphone photos, documents are increasingly being transmitted as images. We do research to ensure perfect digitization.
- Automatic cropping of images
- Automatic alignment of images
- Automatic perspective correction
- Superresolution for documents for better readability
- Handwriting recognition
- Realistic text removal
- Font recognition
Business management
Whether it is incoming mail management, document processing or BPM, CIB also conducts research to make everyday work easier.
- Completeness check of documents
- Fraud check
- Predictive BPM
- Best possible MRC compression for each document with automatic quality control
- Traceability of AI decisions
- Document classification
- Address field correction for postage optimization
- Automatic text suggestions and alternative formulations
- Intelligent email mailroom
- Virtual office robot
Always learning!
Continuous Learning
An AI can learn from mistakes. Corrections adjust the AI's training, and the new AI replaces the old one.
Federated Learning - Data protection matters
To protect critical data, we rely on decentralized AI training: data worth protecting stays in place, only the AI's learned knowledge is shared!
Your own AI
Your specific use case is not yet supported by AI? Our AIs can be retrained. Talk to us!
Mobile AI
AI on the go? With today's computing power and the know-how of CIB no problem! Feel free to contact us about this possibility!
AI training for a good cause!
CIB crowdsource
Through our crowdsourcing platform, anyone can help train our AI models.
We take privacy very seriously: all data processed on the platform by the hardworking members of our community is irrevocably anonymized or was already free of real personal data.
For a good cause!
Of course, diligent help is also rewarded: For each task, you will receive points that can be donated to certified charities.
Register now and join in!
