The problem
You paste the same lecture notes into ChatGPT before every study session. The chat gets long, the answers drift, and next week you paste them again. Worse, the answers lean on the model's general knowledge, not on your course. When the exam follows your lecturer's framing, that gap costs you marks.
How students use Hjarni
Keep one folder per course. Inside it: your lecture notes, the key concepts in your own words, practice problems, and the questions you want to answer on exam day. Connect Hjarni to ChatGPT or Claude once. From then on, any chat can read that folder. You never paste your notes in again.
A typical course folder
- Modules: one note per lecture, with what was covered and what the lecturer stressed
- Key Concepts: one idea per note, explained in your own words
- Exercises: practice problems with your worked solutions
- Retention Questions: the questions you want to answer without looking, each with a short answer
- References: readings, slides, and recordings worth going back to
- AI instructions: "Only quiz me from my notes. Tell me which note an answer came from."
A Key Concept note is short and written by you:
# Opportunity cost
Definition in my own words: what you give up by picking
one option over the next best one.
Why it matters: every choice in the supply unit is
framed this way in the exam.
An example: studying tonight costs the shift I could
have worked.
A common misconception: it is not the money spent.A Retention Question is even shorter:
# What is the opportunity cost of a free lecture?
Answer (short): the next best use of that hour.
Source concept: [[Opportunity cost]]
Last reviewed: 14 SeptemberA concrete workflow
Sunday, week six of Microeconomics. You open ChatGPT on your phone and type: "Quiz me on the five oldest questions in my Micro retention folder." ChatGPT reads your Retention Questions folder, asks the first one, and checks your answer against the one you wrote. Miss it, and ask it to open the matching Key Concept note and walk you back through your own explanation. Tell it you got one right, and the template asks it to update the Last reviewed date, so the oldest questions come up first next Sunday.
Two days before the exam, ask: "Write five practice questions on module 5, using only my notes." It reads your Module 5 note and the concepts linked to it. Ask it to name the note behind each question, so you can check it.
New lecture? Add a note, and it is there for the next study session, in ChatGPT and in Claude.
Start with the Course Notes template
The Course Notes template sets this up for you: the five folders, a starter note in each, and folder instructions that ask the AI to quiz you only from your own material. Paste the template link into Claude or ChatGPT and it builds the structure for you.
Why not a Project or a flashcard app?
ChatGPT Projects and Claude Projects keep files inside one assistant. Your notes then live in that one app, and a classmate on the other assistant cannot read them. Hjarni is one knowledge base that ChatGPT and Claude both read, and you can share a course folder with your study group.
Anki is built for spaced repetition and does it well. Hjarni is lighter and keeps each question next to the note that explains it, so the AI can teach you the answer from your own words. You can use both.
An AI that has read your lectures can help you study. One that hasn't is guessing what your exam will ask.
What students keep in Hjarni
- Lecture notes: one note per lecture, searchable across every course
- Key concepts: your own explanations, which the AI is told not to rewrite
- Worked exercises: problems and the solutions you wrote
- Exam questions: short questions with short answers, reviewed oldest first
- Reading lists: sources with a line on why each one matters
- Study group folders: one shared set of notes every member's AI reads
Writing a thesis or a dissertation? That is research work. See Hjarni for Researchers.