If the chemical industry of the XIXth and XXth century owes its evolution to empiricism, that of the XXIst will draw it from data and Artificial Intelligence: molecular innovation is a field in the process of being totally transformed by the deep learning.

If, for example, you are looking to produce new chemical compounds, the number of candidate molecules to be tested is quite simply astronomical: your experimental quest is therefore akin to trial and error on a case-by-case basis. Even relying on complex simulation software, the approach remains very risky: today the discovery of molecules with precise characteristics is a complex, long and expensive process. So what do we do ?


What if it was possible to train artificial intelligence to predict chemical structures with a specific property? What if deep learning methods applied to chemistry could be used to accelerate the discovery and selection of high performance molecules?

AlchemAI from Alysophil is an innovative molecule design application, a revolutionary tool which foreshadows a new shift in industrial chemistry: faster discovery of new molecules meeting particular specifications; enough to spare the industry laborious and costly iterations!


Alysophil has a library of one million molecules with their characteristics, never synthesized, built by several deep neural networks, trained since 2017. The creation of this database is the result of deep learning technologies. This exceptional base, crossed with public bases, will allow our customers to create molecules with the desired characteristics in a short cycle.


Our model makes it possible to explore - in silico - large chemical spaces that would be prohibitively expensive with traditional experimental approaches, and to solve complex problems for markets with very high demands in terms of innovation: defense, aeronautics, spatial, or

cosmetics, aromas and perfumes, phytosanitary ...




Imagine an application that allows you to
reduce your experience time

and your research costs up to 50%






By generating a list
of molecules under constraints
that meet your needs

AlchemAI offers you concrete technological solutions



at different levels of the value chain:

research, modeling, production.

- in the form of a secure application with user license,

specially developed for your specific needs
by industrialists, for industrialists.


- in the form of operational results, generated by us.


With AlchemAI, everything becomes possible: move the cursor on creativity, innovation and time to market



Deep Learning is one of the main technologies of Artificial Intelligence ; it is a derivative of “Machine Learning” : the machine is able to learn by itself, unlike programming where it is content to execute literally predetermined rules. Deep learning uses deep neural networks to identify structures in huge volumes of data. In computer science, "neural networks" are sets of algorithms modeled on the biological structure of the human brain.  These are biomimetic technologies.

The possibilities are endless today and deep learning technology helps us to achieve new technological advances, real shifts, such as here with AlchemAI.




1 million molecules never synthesized are available today!

Alysophil announces the generation of a library of

1 million molecules never synthesized, built by several deep neural networks, trained since 2017. The creation of this database is part of an approach of interconnection of deep learning and a deeptech industrial activity. This exceptional database will meet the complex specifications of our customers and produce the corresponding molecules on the basis of a short cycle using new syntheses made possible by our flow chemistry technology.