Every studied course produces the same artifact: a pile of watched lectures and a growing list of questions that fall between them. "Didn't she define this differently in week three?" "Which lecture had the worked example with the matrices?" "How does this week's concept connect to the thing from the beginning?" The answers exist — you watched them — but retrieving a moment from twenty hours of video is archaeology, so the questions pile up unanswered until the exam makes them urgent.

Office hours exist for exactly these questions, and office hours are two hours a week with a queue. What you want is the course itself, answerable: your lectures as something you can ask.

deep_context makes the pile conversational: scope a question to your imported course lectures and it answers from what the lecturer actually said, citing the exact chunks — so "where was that worked example?" returns lecture six, minute thirty-four, with the explanation summarized and the source one click away for the rewatch.

library_search and library_list are the catalog: find the right lectures by title and tag instantly (free), which is how questions get scoped to the relevant weeks. Cross-lecture questions — the connective tissue courses are bad at teaching — become askable: "how does week eight's method relate to week two's foundations?" answered across both, cited to both. The course stops being a pile and starts being an office hour with no queue.