Bingo. Dismissing Khan Academy because it's currently mostly known for videos is just as dangerous as dismissing any tiny HN startup that's focusing on a core idea and, once lessons are learned, iterating from there.
We're not arrogant enough to think we'll start out by building a generic AI solution to education.
We do know that we have a) a lot of video content that students are constantly watching, b) dynamically generated math exercises that students use to practice every day, and c) some very encouraging in-school pilot programs.
We're using those experiences to learn and expand. The YouTube video linked in the article shows the edges of our progress so far.