Thank for your for sharing this video Bobur about LLM App. It is a great application given the project just started recently. This is great and like you pointed in the slides, you don't need complex framework/tools (vector db's etc). One small comment/feedback. In the discount app architecture diagram shown around 3:35, it will be nice to number the arrows, based on a user query and step of the operations. KNN is k-nearest neighbors, and ANN is approximate nearest neighbor, techniques used to search indexes based in similar search.