At the heart of Anzyz, we find the self-learning algorithm, Anzyz AI-Insight or Corpus Cube Linguistics - CCL™. This Natural Language Processing (NLP) technology is the platform which generates all our current and future products. Anzyz AI-Insight is capable of reading and comprehending both unstructured and structured digital text and numbers. Because the algorithm is self-learning, it does not require word packages or tagging, which eliminates a lot of manual labor. Furthermore, it can understand irony, sarcasm, double-meanings, and emoticons.

The Anzyz algorithm can comprehend any language, as it is language neutral, and can even understand dialects and misspellings. In addition to this, the technology can be used on both big data and small data.

To summarize; Anzyz AI-Insight differs from other, similar NLP-technologies due to:
• There is no manual tagging required.
• The system is language neutral.
• It draws on associations, due to a profound syntactic understanding.
• It delivers accuracy rates above 90%.

Anzyz’s Natural language understanding technology is built from the ground up on fundamentally different principles than existing state-of-the-art technologies. It is based on three great inventions by Professor Ole-Christoffer Granmo. The inventions have never been or will never be published.

The Anzyz technology is built from the ground up on fundamentally different principles than existing NLP/NLU technologies. Anzyz uniquely integrates 3 different machine learning approaches; supervised, unsupervised and manual rule-based learning.

Unsupervised training; train on any set of raw (untagged) textual data at limited resources and with limited computational efforts. The technology understands interdependencies and connections in words, sentences and context – wholly based on self-learning from a raw data set.

The result has a high precision, even when trained on limited data sets, which demonstrates superior practical value versus existing technologies in a broad range of use cases. Furthermore, the Anzyz technology understands nuances and context for any natural language, including e.g. jargon, slang, non-Western language and more.

Identifying allergy information in health records at Sørlandets Hospital

  • Research project in collaboration with Sørlandet Hospital Trust (SSHF) in 2017
  • Objective is to identify information relevant to patient allergies from health records by the use of software/machine learning techniques
  • Anzyz significantly outperformed all conventional techniques

Instead of building a global concept model, Anzyz retains all the detailed linguistic textual fragments. The complex probabilistic network of relations ensures that no relevant information is discarded. Word embedding models are formed by relating critical knowledge base fragments at query time. Other algorithms cannot model the linguistic nuances with the same complexity as Anzyz.

It's unique qualities are summarized below:

Fast, Language Neutral, Accurate

Description of the production of the product / service

The actual production of the product and service can be divided into three steps:

1. Data is uploaded
Documents in any language or dialect, and in any file format, are processed efficiently with Anzyz AI-Insight. These are either big- or small data sets, provided by the customer.

2. AI-base
Start analyzing your data! Real-time insights and analytic tools allow you to work live with the system to gather concrete information from the AI-Base. The interface is user- friendly and can be tailormade to fit your specific needs.

3. Analysis
Our self-learning technology automatically transforms the uploaded data into an AI-base.


While developing a Clinical Decision Support System (CDSS) at Sørlandet Hospital, the AI technology of Anzyz was tested against Facebook Starspace (2018). The aim of the research was to develop a system for identifying and classifying allergies of concern for anesthesia during surgery. The aim was to provide healthcare professionals with a clinical decision-making tool to improve their clinical workflow and the patient-clinician communication, as well as increase patient safety. Today,the CDSS is used every day at the hospital and healthcare personnel report great satisfaction with the system. The comparison with Facebook Starspace, can be see below:

Mastering a Vocabulary

800 000 Electronic Health Records

Anzyz CCL Facebook Starspace
Captures the meaning of 1.8 million phrases Captures the meaning of 44,000 phrases
Multi-word phrases Single word phrases
Two days to learn the meaning of 1.8 million phrases Ten days to learn the meaning of 44,000 phrases


Anzyz co-founder, Ole-Christoffer Granmo, and House of CAIR are placing Norway on the map with a brand new deep learning supercomputer!