Nkuzi

Nkuzi is a study partner for medical students who learn by explaining out loud. You hand it your lecture slides as PowerPoint, PDF or Word, then talk through the topic. As you speak it ticks off each point on a checklist built from those slides, so you can see at a glance what you have covered and what you have missed.

When you are done explaining, two buttons do the heavy lifting. “What did I miss?” lists every point you skipped, and “Check me” compares what you said against the slides and flags anything that contradicts them — and every correction quotes the exact slide line it is based on, so you can tell a real mistake from a mishearing. At the end you get a short recap you can paste straight into WhatsApp for your study group.

The whole thing runs on one old laptop with no internet and no subscription. The author built it for friends studying medicine in Nigeria, where data is expensive and the network drops out, so speech recognition, the checklist and the corrections all happen locally on a 2014 laptop with no graphics card. Your voice and your slides never leave that machine, and on GitHub the project has no stars yet.

Honest limits: there is no hosted demo to click through, so you run it yourself from the repository with Python, Node and Ollama. Transcription is the weak spot on Nigerian-accented English and medical terms, “Check me” takes one to two minutes on the old laptop, ticking follows the topic rather than whether you were right, and only one person can explain at a time. Old .ppt files and scanned PDFs need to be re-saved first.