Khanmigo Upgrade Pathway

I have been hearing about how Khanmigo has done wonders for educators, but students have not been getting nearly the same benefits out of its prompts. From my experience, Activities like the debate section have been super-engaging and helped me think more deeply about it's topics. However, Khanmigo as featured in videos have been a bit lackluster and more as a review tool.
What I have found in using other AI tools is that those with more context of material outside of that just from the course material helps me find more "Aha!" moments where material from one place is helping me appreciate the new material I am learning, even if it is from the more basic ones. Just now, I went through Gemini Notebook's featured notebook about philosophy and after a few prompts, I was shocked to find out Plato's Theory of Forms was INSPIRED by the Pythagoreans, showing just how much trying to prove the Pythagorean method truly helps unlock deeper thinking skills, as it did with the OG philosophers of the West.
Similarly, after looking back at a video about the quadratic formula, something new clicked when Sal's video on it brought up the formula is used to find the zeros of a quadratic function.  The fact that the ZEROS were the main thing the Quadratic formula was being used for made me recall how critical zeros were in calculus. (After more reviewing, I found out that the zeros of a derivative were "critical points" that locates local maxima.) SUPER important connection, and I know LLMs can be another great tool to help students find that Aha! moment, especially when their lecturers may not be so good at revealing these connections.
I want to ask then of the Khanmigo engineers to see if they can expand the context window of Khanmigo itself so that it can use the ENTIRE Khan Academy library as a sort of RAG to find these connections across subjects and units within a subject. It really unlocks a whole other level of appreciation of knowledge when it can help find these connections. After all, it's become an expectation of LLMs and AI to be really good at finding patterns and distinctive material for a solution.