 Hello, I'm Mathieu Donorin from ST Microelectronics. Hi, I'm François Drogebois from Cartisium. Cartisium is a software company that worked in Toulon, in France, and we have offices in Paris and New York. We are specialized in artificial antigens at the edge. To be more precise, machine learning running directly on microcontrollers such as the ACM32. We developed an innovative solution called NanoHAI that enables real machine learning at the edge, meaning analysis but also learning directly on the microcontroller, and that is unique on the market at the moment. Yeah, sure, we have a sensor tile board and a brushless motor, so everything needed to demonstrate the power of NanoHAI running on the STM32 microcontrollers. The sensor tile is a tiny IoT module that packs powerful processing capabilities. The veraging on the 80 MHz STM-L4 microcontroller, Bluetooth low-energy connectivity, as well as a wide spectrum of motion and environmental main sensors, including a digital microphone. It's an ideal platform to prototype and develop AI at the edge. The demonstration is around predictive maintenance. The overall idea is to use the accelerometer and the STM32 microcontroller with NanoHAI. First, we will learn the vibratory environment of the motor and then we will introduce anomalies. The demo can demonstrate the ease of learning of the vibratory signal directly in the STM32 microcontroller and once the training is complete, its analysis to detect the default also run inside the microcontroller. In terms of possibility, the scale is the limit, but here are two ideas that stick to reuse case. An ultrasonic microphone connected to an STM32 microcontroller, running NanoHAI, learn the usual ambient noise of a place inside an electrical cabinet, for example, and then looks for the appearance of electrical arc to prevent fire and explosions. It could also work extremely well for analysis of energy, water, electricity, gas. In this case, NanoHAI learned the consumption patterns of a room and then detects any unusual consumption due to a leak, a tap left, open, same idea for gas, et cetera. As you can see, ST and Cartesium are very complementary to bring AI at the edge, either for unsupervised or supervised machine learning approach. We are collaborating together to provide the best solution to our customers, either in terms of performance, low power, ease of implementation, or development kits. We are working on specific optimization for NanoHAI on STM32 with this focus on very low power applications. Predictive maintenance is the first use case we are demonstrating together. Cartesium is part of the ST Partner Program, and I invite you to visit st.com slash partner to find out more about this partnership.