HKU Open-Sources DeepTutor, an AI Tutor Built on Your Own Notes

Studying off a pile of PDFs, lecture recordings, and half-finished notes gets messy fast. DeepTutor wants to be the one workspace all of that lives in, and it is free and open source. HKU put it on GitHub, where it has already pulled in 48K stars.

It is not a one-question chat box. The whole idea is that it works from your material and remembers where you stopped, so you are not re-explaining your syllabus every session.

What Happened

DeepTutor is an AI private tutor open-sourced by HKU. You feed it your textbook notes and it teaches from them: it explains, generates practice questions, and draws diagrams, all inside a single workspace instead of scattered across five separate tools.

The part that stands out is memory. Switch sessions and it picks up where you left off, and you can open a three-layer breakdown of your learning records to see how far you have come and where you got stuck.

It also handles online courses. Paste in a Bilibili or YouTube link and you can ask about any second of it you did not follow, so a two-hour lecture becomes something you can actually interrogate.

Setup is short by the project's own account: three lines of code. If you would rather not configure anything, you can hand it to Claude and let it install itself, which is an odd sentence to write about a study tool but there it is.

Why It Matters

Most study chatbots forget you the moment you close the tab and know nothing about your course. DeepTutor is built around the opposite assumption: your notes are the curriculum, and the tutor should keep the thread across sessions.

For anyone working through recorded lectures, that video feature alone is worth a look. Asking about the exact minute something went over your head beats rewatching the whole thing at 1.5x. It is open source, so there is no subscription sitting between you and your own notes.