The starting point of PI is a unique method in the field of machine learning which overcomes known problems of artificial neural networks. From the very beginning there were inquiries from large German engineering companies for development projects, and we joined several research projects investigating artificial intelligence in medicine, chemistry, and aviation, which kept us busy for some time.
Originally our claim was to offer an "AI for everyone", since using our algorithms requires no prior knowledge of machine learning techniques. In the software package STOCHOS all methods are combined in a universally applicable tool, and unlike neural networks, they do not have to be adapted to every new problem.
Then came a chicken-and-egg problem. Our contact persons in the companies evaluating STOCHOS usually came from classic neural networks and often had no interest in introducing other methods. An exception from the very beginning was Bosch, which had begun investing broadly in artificial intelligence at a very early stage.
The leap through CADFEM
At the Ansys WOST conference at the end of June 2023 in Weimar, CADFEM became aware of us in their search for partners in the field of artificial intelligence. After that, everything happened very quickly.
- End of 2023. We became a new member of the group, and thanks to CADFEM's long-standing customer relationships we now talk directly to the actual potential users of AI inside the companies.
- April 2024, CADFEM Conference Darmstadt. Since then we have spoken with engineering and AI specialists from international market leaders on an equal footing about putting ML into work processes.
- Since the conference. A whole series of PoCs on simulation data, experimental data, and combinations of both, in component optimization, process control, simulations in chemistry, and other applications.
The feedback has been overwhelming, perhaps because many leading companies that already simulate in development had tested several approaches during the AI hype and could quickly grasp the advantages of STOCHOS. What we hear consistently is that implementation is easy, so partners can use the methods on their own for further internal development.
Our international team has been interested in projects outside the DACH region since the beginning. With the many branches of the CADFEM Group behind us, within a few months we started projects with companies in Europe, the USA, India, and Southeast Asia.
Looking back after half a year, we can safely say that thanks to the support of CADFEM we made the leap from a software startup onto the world stage of artificial intelligence in a very short time, and we remain confident that probabilistic methods will continue to prevail.
Nils Odenthal, Managing Director at PI Probaligence GmbH
Read the full journal here: CADFEM Journal 2024
