Smartly Connected. Seamlessly Integrated. Truly Automated.

Our colleague Yini Xing from Research & Development will provide an outlook on the future development of artificial intelligence, classify current advances in model development from both a technological and timeline perspective, and present the AI tool developed by HighPots for the e-commerce sector.

HighPots and BMC demonstrate a cooperation model in which multiple companies jointly organize their AI fine-tuning. Instead of each organization repeatedly performing the same costly training processes, base models and training steps are sensibly distributed. Each company thus only invests in truly value-adding, individual processes and data. The presentation highlights the associated AI processes, cost-sharing models, modern software architectures, and the establishment of an intelligent, distributed AI infrastructure.

Heavier weight, instant torque, range affected by rolling resistance, and—for the first time under Euro 7—regulated wear: Electric vehicles place conflicting demands on tire compounds. The search space is too large for comprehensive test bench work. Our machine learning framework narrows it down before testing begins in the lab—it prioritizes materials and compound formulations based on existing measurement data. Here, we show how it works.

In collaboration with Twerd, HighPots is introducing a smart energy management system integrated directly into the inverter. The embedded AI dynamically allocates PV power among self-consumption, battery storage, e-mobility, local consumers, and the public grid. In doing so, it takes into account forecasted consumption patterns, available storage capacities, and individually defined energy reserves. The solution developed by HighPots transforms the inverter into a smart control center for decentralized energy systems. Based on consumption and generation forecasts, the system continuously determines how to optimally utilize available PV energy. Among other factors, it takes into account the charge levels of building batteries and—via connected wallboxes—vehicle batteries. Individual energy buffers and minimum reserves can be defined for each participant. Excess energy can be automatically distributed to other participants—such as neighboring households or buildings—and then recorded and billed. At the same time, the system determines which amount of energy is stored, used locally, or fed into the public power grid. The solution thus combines embedded AI, energy management, storage control, and local energy distribution in a single system. Embedded AI from HighPots and power electronics from TWERD Energo-Plus for the intelligent control of decentralized energy systems.

HighPots presents a fully on-premises AI solution that helps interdisciplinary research and development teams continuously improve their research methodology. It detects weaknesses in methods, technologies, processes, and collaboration; identifies previously unrecognized connections; and uses these insights to develop concrete suggestions for improvement. To do this, the AI aggregates information and results from a wide variety of sources and considers not only individual research findings but the entire research process and its evolution over time. I consider this wording to be much more accurate because it now highlights the actual product promise: not simply aggregating information, but systematically improving research.