Why do we do research?
In our quest to lead in AI innovation, we not only harness state-of-the-art technologies but also pioneer the development of new ones, venturing into unexplored territories of artificial intelligence. This relentless drive propels us beyond conventional boundaries, enabling us to solve complex challenges more effectively across various domains. Our commitment to staying at the forefront of AI research ensures continuous growth and exploration, as we shape the future of technology and unlock potential yet to be fully understood. Through this approach, we're not just adapting to the evolution of AI; we're defining it.
How we do research?
Our research process is like that of universities or other research institutes. We start by reviewing academic literature and identifying state-of-the-art for our current challenges. We are particularly interested in solutions that show potential to be further developed and scaled to handle more difficult problem instances. With the most promising approaches we run an algorithm engineering loop, where we implement and then repeatedly evaluate, profile, analyze and optimize the algorithms until they are capable of handling real industry scale problems.
Finally, we, on top of implementing and integrating our solutions into our platform and products, also publish them as patents and scientific publications.
What is current focus?
Our current focus is on creating methods to learn action models from logs for enhancing process automation, including software testing. By capturing changes in dynamic systems, we're able to automatically generate models that detail all valid transitions, significantly streamlining tasks such as test generation, planning optimization, and system verification. These efforts lead to the synthesis of symbolic action models, which are instrumental in automating system validation and reducing the associated effort. For more insights and detailed descriptions of our work, explore our research blog posts, patents, and publications through the provided links.
