 Hi, I'm Louis and I'm going to introduce you to one of the ST-AJI solution, Nano-AJI Studio. In that case, we're going to have this demonstration running focused on predictive maintenance at the edge, on microcontroller. This is enabled by Nano-AJI Studio. This is a full software suite that lets you develop machine learning algorithms locally for deployment on microcontrollers. That lets you develop something to do classification, regressions or anomaly detection that we're going to see now. We have here the demo that we're going to present today on predictive maintenance. We have two pumps, two identical pumps, playing to two different circuits. On the first one we can see the Proteus iValboard that's monitoring this pump using the vibration, using axiometer from the vibration generated by this pump. Right now, I'm actually going to start the training of that Proteus on this pump. All of the training is happening inside the Proteus. There is no data being sent back, there is no data being sent to a centralized cloud or to any other solution. We are just going to be seeing on the screen displayed on your screen what's happening reported by Bluetooth. This is just so that we can see what's happening and what the Proteus is telling us is happening. So now that the learning is over, we can see that the status is normal, similarity is very high. This means that it's very similar to what it has learned. Now if I shut off one of the valves of the water circuit, we can see that the status goes to anomaly and the number of similarity goes very low, it's very different from what we've learned. If I partially open it, we have that case where similarity is higher than it was when it was completely shut, but it's still anomaly. So now, if I move this board to the second pump, it's an identical pump, it's just plugged into a different circuit. We are detecting this pump as an anomaly as well. The problem or benefits with that really is that we have algorithm that need to generalize to different environments. If you have thought about everything during your development, that's great. Most times you're going to have missed some cases or edge cases that are going to appear once deployed in the field. And now with Nano-HGI Studio, we're resetting the knowledge and starting a new learning cycle, you can actually learn on the board itself. Now that we've finished the second learning on this pump, we are detecting this pump as nominal again. We have now learned what the pattern of this pump looked like. It's the same pump but plugged into a different circuit which basically emulated different environments for our device, in that case, pump to be in. We can now monitor it natively. It's the same algorithm, it's the same board, it just learned a new environment. And obviously, if I create anomaly by shutting off one side of the circuit, we're detecting this as an anomaly as well. Thank you very much for your time and attention. This is a demonstration of one of the algorithms by Nano-HGI Studio. We have other algorithms, other use cases. In general, it's one of two of SD solutions with QBI that lets you fully implement edge AI solutions for anything going from predictive maintenance to general monitoring of your solution and qualification of errors. You can find the link for on the demo and on Nano-HGI Studio in the video description. Thank you very much.