Feature Suggestions for Enhancing Assistant Performance on Dust

Hey! I’m suggesting two features that I’m currently building on my own because I really need it! (but would be great to have it directly on dust) 1- Having a button who can take all the feedback you’ve made to an assistant and suggest you a new prompt who’s improving the assistant 2- Having the possibility to create evaluation dataset (manuel label) and run different version of your assistant on the dataset so you can improve his performance and keep the best prompt/model based on quantitative metrics