If I could give every student one piece of evidence-based study advice, it'd be: quiz yourself.
Not re-read your notes. Not highlight. Not make a mind map. Quiz yourself. Cold. Without looking. The thing the cognitive science literature has been screaming about for thirty years is active recall, and the gap between how effective it is and how rarely students actually do it is enormous.
There's a reason for the gap. Making your own quizzes is annoying. You have to find the material, write the questions, write plausible wrong answers if it's MCQ, hide the answer key, grade yourself honestly. By the time you've done all that, you could have just re-read the chapter twice. So most students never do it.
Fennie's quizzes are the lazy fix. Two clicks from any note or document. Auto-graded. The score feeds Memory. Tomorrow's plan adjusts.
This is the part of the system I'd argue is the highest-leverage. If you do nothing else, do the quizzes.
The two-click flow
Open a note. Or a document you uploaded. Or a chapter inside one.
Click Quiz. Pick how many questions (default 8) and what format (default mixed). Click Generate.
You get a quiz scoped to that material. Take it now or add it to tomorrow's plan. When you submit, you get a score, per-question feedback, and explanations on anything you missed. The whole result feeds Memory.
That's the whole flow. The barrier between "I should quiz myself on this chapter" and "I am quizzing myself on this chapter" is about 15 seconds.
Quiz formats and when to use them
Fennie supports a few formats. The right format depends on what you're studying for.
Multiple choice
Pattern recognition. Useful when the test you're prepping for is MCQ — MCAT, USMLE Step 1, AP exams, AAMC sample questions, most undergrad intro courses. The wrong answers Fennie generates aren't filler; they're plausible distractors based on common student misconceptions. That's where MCQ practice gets its value — you have to know the material well enough to eliminate wrong answers, not just recognize the right one.
The risk with MCQ is you get good at recognition without being able to produce. Pair MCQ with at least one other format if your test isn't purely multiple choice.
Short answer
Production. You write a sentence or two. Fennie grades against the key concepts that should appear, not the exact wording. Useful for intermediate-stakes topics where you need to know the material well enough to articulate it but not write a full essay.
This is the most under-used quiz format. Most students don't reach for it because writing answers takes longer. It's also where you find out you don't actually understand something. Highly recommended for hard sciences, technical material, and any course where the test asks you to explain.
Free response
Longer-form. A paragraph or more. Fennie grades qualitatively — what's there, what's missing, what's wrong. Use for essay-based subjects, case briefs, longer explanation problems.
The grading on free response isn't pass/fail. You get a rubric-style breakdown so you can see what would have gotten partial credit on a real test.
Fill in the blank
Procedural. Equations, statutes, mechanism steps. Better than MCQ for material where the specific token matters.
The mistake most students make
Treating quizzes as a thing to do only before tests.
The way most students use practice quizzes: ignore them all term, find a question bank the week of the midterm, hammer it for two days, take the test. This is fine for raw cramming. It's terrible for retention and it's worse for actually learning the material.
The way Fennie wants you to use quizzes: as you go.
End of a lecture, you have notes. Generate an 8-question quiz on those notes. Take it that evening or tomorrow morning. Find out which two concepts you didn't actually understand. Re-read those parts. The whole loop is twenty minutes and it locks in material that would have otherwise leaked out by next week.
This is what active recall actually is in practice. Not a wall of practice questions before the exam. A small daily test on what you just learned, repeated.
How Memory uses your quiz scores
Every question you answer goes to Memory tagged with the underlying concept. Not just "got it right" — but "got the chain rule in this specific application right" or "missed res ipsa loquitur for the second time."
Memory uses that picture to make three decisions:
What goes in tomorrow's plan. Concepts you've quizzed solid on drop out. Concepts you've missed twice come back, often in a different form (a flashcard, a different quiz angle, a chat suggestion to walk through it).
What flashcards surface. A concept you missed on a quiz might come back as flashcards over the next few days, not just another quiz immediately.
Which lessons or chat prompts get suggested. If you're consistently weak on a topic, chat will offer to walk through it next time you open it.
The whole thing is supposed to feel quiet. You take a quiz. You see your score. The plan tomorrow is slightly different. You don't have to think about what to study next.
Test-aware adaptation
Add a midterm to your calendar. Watch what changes.
The week before the test, the daily plan rebalances toward that course. Quiz frequency on that course goes up. Quizzes start pulling from earlier material, not just the most recent week — because by midterm time, the question is whether you remember unit one, not whether you remember last Friday's lecture.
The day before the test, the plan often suggests a cumulative quiz. Maybe 20 questions across all the units the test covers. Take it cold. Where you score well, you can stop worrying. Where you score badly, that's your evening's review.
Day of the test, the plan eases up. A short refresh on the weakest topics, not a wall of new material. The morning of an exam isn't the time to learn something — it's the time to make sure your hands know where things are.
This sequence — 4 days out, 2 days out, day before, morning of — happens automatically when a test is on the calendar. You don't configure it. You just put the date in.
Pre-test quiz patterns by class type
A few examples of how this plays out for different students.
MCAT, four days from a full-length
Daily plans get heavy on cumulative quizzes across the section weights. Bio/biochem each get a quiz day. Phys/chem combined. Psych/soc. Memory's picture of your weak topics drives which subtopics get more questions.
Two days out, plan shifts to mixed-section quizzes — questions stop being clustered by topic, mimicking the real exam's randomness.
Day before, a focused review of the topics you've consistently missed. Light volume.
Test day morning, almost nothing. Short flashcard review. Maybe a few comfort questions.
Law school first-year, three days from a 1L final
Issue-spotter style quizzes. The cumulative quizzes go long — fact patterns with multiple issues, like a real exam. Memory has been tracking which doctrines you've been shaky on; those show up disproportionately.
Two days out, you'd typically generate a free-response quiz from your outline and write actual paragraphs. This is the format closest to the test. Fennie grades the answer against the doctrines that should appear.
Day before, light. Re-read the two sections of your outline that the quiz scores say are weakest. Sleep.
AP Calc BC, week before the exam
Mixed format quizzes. MCQ for the multiple-choice section practice, free response for the FRQ section. Memory's been watching all year — the weak topics in October are usually still your weak topics in May, and you'll notice them in the question distribution now.
The week ramps up to a full-format mock on the weekend. Take it timed. Score it. The Monday plan reflects exactly what fell apart.
Honest weaknesses
Some quiz formats are better than others, and Fennie won't pretend otherwise.
Auto-graded free response on highly subjective material is decent but not perfect. Fennie can tell whether your answer hits the key concepts. It's worse at evaluating writing quality, novel arguments, or unusual phrasings. For a real essay-graded class, you should be supplementing with chat critique on a few sample answers, not relying purely on free-response quizzes.
Question variety on small notes. A 3-line note will not generate a great quiz. Quiz quality scales with the source material. Pull from longer notes, full chapters, or whole units when you want a quiz that actually probes you.
Computational problems with multiple solution paths. Fennie can grade whether you arrived at the right answer. It's less good at grading partial credit on a wrong answer with a correct method. For Statics, Diff Eq, multivariable calc — pair the quizzes with chat for working through the problem.
The minimum viable habit
If you're skeptical and want to try one thing this week:
After your hardest lecture, generate a quiz on that lecture's notes. 8 questions. Take it the next morning before the lecture meets again. Look at what you missed. That's the minimum. Twenty minutes.
Do that for two weeks and your quiz scores in that class will measurably go up. Memory will start surfacing the right material in your daily plan because it has real data on you. The whole system gets better the more you quiz.
The students who plateau on Fennie are the ones who skip quizzes. The students who get the most out of it are the ones who took ten minutes a day for active recall and let the rest of the loop close around them.
Try generating your first quiz from a note. Start with Fennie free.