Intrinsically efficient as the cortex

Instead of considering the human brain at a microscopic/neuron level as in deep learning, we replicate the brain’s behavior at a more macroscopic scale where large neuronal populations have a dedicated function like the perception of motion or curvature. 

Generic to all sensing modalities 

Our technology is generic to all sensory perceptions and can thus treat vision as well as smell, sound… like the cortex does.

Unsupervised learning
Like a child, the technology extracts the underlying laws and patterns from its environment and is able to use them to perform high-level tasks.

Incremental learning
For each new task, instead of starting from scratch Organic AI draws links from previous learning and builds an ever-growing knowledge of the world, without jeopardizing previous behaviors.

Ultra low data
Our technology requires very little data to perform its tasks, although it can scale up to large databases.

Ultra low energy
Double energy efficiency : our models' efficient training on limited data samples reduces the need for computing power. Our algorithms are also considerably less computationally intensive than deep learning technologies. 

Ultra low cost
Integrated on a single chip, our technology can easily be installed on all types of devices, allowing you to build ultra low cost AI solutions. 

Protecting privacy by design
Our GDPR-compliant AI technology can be run and updated locally and in real-time. It does not need connections to any cloud platform to feed and/or control it. It prevents "Big Brother” solutions and potential hacking through Internet connections.

Explainable by design
Unlike deep learning’s “black box” AI solutions taking decisions which cannot be tracked or explained, our technology gives the possibility to audit its processes and to know how and why. We believe in transparent algorithms. As an aside the technology is easily certifiable where it matters (e.g. autonomous self-driving).