Building an AI Tutor With Gemini-CLI and Obsidian
Virtual
An explanation and a technical demo of how to build an AI tutor for any skill with Google’s Gemini-CLI and the Obsidian text editor. By using Gemini in the filesystem, users are able to save prompts from the LLM as files and then edit those files directly. This allows anyone to use Gemini-CLI to create a local tutor that will quiz you, grade quizzes, have persistent storage in the filesystem, provide links to relevant documentation, and iteratively improve as the user uses the tutor system. Because it is saved in the filesystem, a user can come back to the folder days or weeks later, and the full AI context will still be there, by utilizing the GEMINI.md system instructions file as persistent state storage. I will have some slides to explain the theory and setup with a live technical demo creating an AI tutor for a new skill in real time, showing how it can be designed, implemented, and used efficiently and safely.