The future of AI is here.

Quantum-inspired AI
with a shared knowledge base. is built on an artificial neural network, that is to say, a mathematical model consisting of quantum-inspired artificial ‘neurons’, the qubit, which are able to automatically improve their performances through supervised and unsupervised learning. The shared knowledge base is composed of an extremely innovative data network entailing graphs. Within this structure, the nodes are the concepts/subjects/entities semantically connected with the relationships, which become increasingly more sophisticated during the interactions with the users.

There are three main types of nodes.


The lemma, which contains basic pieces of information (e.g., person, cat, vehicle).


The concept, which includes different lemmas, hence creating a hierarchically higher node (e.g., protected animal).


An inference includes both lemma and concept nodes allowing for the creation of deductions, which are the result of a deductive process.

Latest Award.

The Global Annual Achievement Awards for Artificial Intelligence Winner of the Best New Topology for AI.

Other Awards.


1° award Confindustria in 2014.


1° award at Microsoft MVP Global Summit in 2014.


1° award Besta Case at Microsoft Future Decoded in 2015. is able to combine data-based with algorithm-based structures.

leading to clear advantages when compared to the standard solutions, such as:

Quicker training time.

Extremely quick training times due to the fact that, as it works by concepts, huge amounts of data are no longer required (especially with training through large quantity of examples), so that the links between concepts are created only with a single training case. Training through sheer quantities of examples, typical of standard solutions, is carried out with many examples questions and answers, involving concepts and periphrasis. These, despite being extremely similar, aim to find a correspondence between the questions and the examples. is instead based on concepts, therefore requiring only a single training case.

Better conversational performances.

Better performances in the conversational applications, resulting from the usage of different shared knowledge bases which are daily updated and enriched, entailing over than a millions of nodes and relationships, available to all customers and linked to the common knowledge.

Higher precision.

A higher degree of precision and accuracy with respect to the standard neural networks.’s neural network is in fact dynamic, as nodes, concepts and inferences can be added or removed without detriment to the mathematical model and with no need for a new training process.

Scalable costs.

As a result, when compared to the standard solutions available in the market,’s concrete advantages entail a considerably higher degree of precision in terms of output quality, model dynamic and training time saving. In addition, its flexible technology implies a considerable decrease in costs producing evident scalability, which, according to the request rise, adjusts automatically to the server characteristics and the hardware upgrades.

Limited hardware resources.

As it is graph-based,’s technology only responds with the involved parts of the neural networks according to the entered input, drastically accelerating the entire process with respect to standard networks. Therefore, the high-performance hardware resources significantly reduce the costs.