 we're doing a fuzzy logic application in Flask. We have code we generated in the past, but that code has quite a few issues. I will share the code with you in a bit. And we either have to fix it or produce prompts as to how to generate a similar code. And we will start generating this code from scratch. So GPT is keen to help. That's great. Once you generate specific prompts, we also have access to GitHub Copilot. So we might be using that as well. Okay, so we have the HTML. Great. I see the HTML code for your fuzzy logic waveform detector. Yeah, let's run it, run the Flask application, can show it to the bot as well. It doesn't fit the screen. This is currently what the front panel looks like. There are a couple of issues with it. If basic things like all the charts do not fit the one screen. And more importantly, that the detection of the fuzzy logic is incorrect, we would like an option to improve upon it from within the front panel. Keep calling them front panels, because that's what used to be called in LabView. That's LoveView, LoveView, LabView, LabView. Yeah, we have CSS. We can deal with it later. But yes, we do like to review and refine the membership function parameters and modify the fuzzy rules. The noise addition, no, make sure it's not distorting the signal too much. User feedback is a nice idea. But then we can just in the code have a calculation for correct responses, which can be easily tested by looking at the name of the input signal and matching it with the output. And note about the input and output strings, the input strings have, you can just check in the code. But for example, they have a left trapezoid, right trapezoid and trapezoid, whereas the output is more of a family of the signal, it should be returning trapezoid for all these three inputs. Yeah, I would like to do refinement and modifying fuzzy rules eventually. But we would like a more systematic way of doing it. The current display of the fuzzy logic settings doesn't seem complete. We also have a problem in the present detection error measurement. Another thing is that the output wave called undefined should affect the present unknown in the detection error and the unknown percentage of the time. It doesn't seem to work at the moment. We might also want to adjust the names used unknown instead of undefined as the output waveform. They can generate the code or we're getting GitHub co-pilot to do it. Good luck with your refinements. Good luck with your refinements. So it's telling me what to do obviously. It seems like it doesn't plan to generate any code. That's unfortunate. We might force you to use co-pilot. Yeah, that seemed to be a simple task for co-pilot. I just realized that the undefined is actually coming from the code, from the execution of the code. It's an actual error. It seemed to be an error. But that's what is only happening sometimes. It's not happening for all input waveforms. Actually most input waveforms work okay. Okay, there's actually too many issues with this code. Can we use a different approach? More inclined to restart the whole development from scratch. Could you generate prompts for GitHub co-pilot? So we have, we happy with those as they are, maybe accept the triangles. Triangles should be look more like triangles. Now the noise amplitude need to make sure it's actually going to zero, that the slider goes to zero. Do we need the input sets as well or not? Yes, we'll do error handling. Detaining a robust version control system. Remember to take breaks, stay hydrated, and keep moving.