Study Techniques That Actually Work When AI Does the Organizing

Most "AI study tips" articles miss the point. They tell you to ask ChatGPT to summarize your textbook, then go highlight the summary. That's the same low-yield study habit you already had, just with extra steps.

The techniques that actually work — the ones cognitive scientists have been screaming about for forty years — are mostly boring. They require effort. They feel less productive than rereading. They're a pain to schedule.

That last part is where AI is genuinely useful. Not as a smarter highlighter. As the thing that keeps you doing the unfun, effective work on the right cadence.

Here's what's actually worth doing, and where Fennie's adaptive system fits in.

First, the techniques you should stop using

Highlighting. Rereading. Copying notes from a slide deck onto a different surface and calling it studying.

These feel productive because they're easy. You see your textbook getting more colorful. You finish the chapter. You feel done. But every meta-analysis on this topic — Dunlosky 2013 is the famous one, and the field has only doubled down since — puts these methods near the bottom of the rankings. They produce a sense of fluency without the underlying retrieval ability. You read a passage three times, and the words start to feel familiar. You confuse familiarity for understanding. The exam asks you to produce, not recognize, and you're sunk.

I'd add one more: making elaborately formatted Notion notes. The format-tweaking is procrastination dressed in a productive shirt.

What follows are the four techniques that actually move the needle. Each one is annoying in a specific way, which is part of why they work — and part of why most students don't do them consistently without help.

Technique 1: Retrieval practice

This is the big one. Probably 70% of the gain.

The idea is brutally simple: instead of looking at the material, try to produce it from memory. Close the book. What were the four mechanisms of enzyme inhibition? Don't peek. Just write what you remember. Then check.

The act of pulling information out — even when you fail — strengthens the memory in a way that putting it in again doesn't. There's a counterintuitive finding here that researchers call the testing effect: a student who reads a chapter once and then tests themselves twice will outperform a student who reads it three times. Every time. It's not even close.

The catch is that retrieval practice requires questions. Good ones. Calibrated to what you actually need to know. Most students don't make their own quizzes because writing a good question takes longer than answering one, and you can't really test yourself on material you haven't yet learned.

This is where Fennie's note-to-quiz pipeline does the unglamorous work. You take notes — markdown, LaTeX, whatever you're already doing — and two clicks generate a quiz. Not a recall-the-definition quiz. The kind that asks you to apply the concept, compare it to another, predict what happens if a variable changes. The questions you'd write yourself if you had three extra hours per chapter, which you don't.

A medical student studying cardiac pharmacology shouldn't be making flashcards from scratch on a Tuesday night. They should be retrieving cardiac pharmacology on a Tuesday night.

Technique 2: Spaced repetition

Cousin to retrieval practice, and equally well-supported. The principle: review material right before you'd forget it, not right after you learned it. Each successful retrieval pushes the next review further out. Things you struggle with come back tomorrow. Things you nailed come back in two weeks.

You can do this manually with index cards and a Leitner box. People did it for decades. It works. It's also a part-time job.

The hard part isn't the algorithm — Anki has solved that since 2008. The hard part is keeping a deck alive across a 14-week semester while you're also writing essays and going to lab. The students who actually use Anki are a self-selected slice of the medical school applicant pool. Everyone else opens it once, makes 40 cards, and never returns.

Fennie's flashcards are spaced under the hood. You generate them from notes (same two-click flow), and the daily plan surfaces the right cards on the right day. You don't think about scheduling. You don't decide which deck to do. You open the morning plan, and "Review 18 flashcards from Mod 3" is item two of four.

The whole point of spaced repetition is that it works while you forget about it.

Technique 3: Interleaving

Here's the contrarian one. Most students study in blocks. Monday: organic chem. Tuesday: physics. Wednesday: organic chem again. Within a chem session: thirty problems on SN1 reactions, then thirty on SN2.

The research says this is worse than mixing. A study session where you alternate SN1, SN2, E1, E2 problems — even though it feels harder, even though you'll get more wrong in the moment — produces dramatically better performance two weeks later. Your brain has to keep deciding which mechanism applies, and that discrimination is most of the skill you're trying to build.

Same with cross-course interleaving. Studying contracts for two hours, then torts for an hour, then back to contracts is harder in the moment and better for the exam.

This is one of the things Fennie's daily plan quietly does. The morning plan isn't "spend 90 minutes on whatever's most overdue." It's a 3–4 item mix, calibrated to what's coming up on the calendar and what your memory shows as weak. Constitutional law, then a quiz on civ pro, then a flashcard pass on evidence. Built fresh each morning.

You're not supposed to enjoy the switching. You're supposed to learn from it.

Technique 4: The Feynman technique

Named for Richard Feynman, who allegedly used it. The protocol:

Pick a concept.
Try to explain it as if to a curious 12-year-old. Out loud, on paper, doesn't matter.
Notice every place you stumbled, hand-waved, or reached for jargon.
Go back to the source and fix exactly those spots.
Try again.

It's the most effective way I know to find the holes in your own understanding. The places you stutter are the places you don't actually know — even if you can pick the right answer on a multiple-choice question.

The Feynman technique pairs unusually well with the right kind of AI tutor, and unusually badly with the wrong kind.

The wrong kind: ChatGPT, used the way most students use it. You paste in a topic, it spits out an explanation, you nod, you move on. You weren't the one explaining. The AI was. Feynman in reverse.

The right kind is what Fennie's chat is built around. Paste a problem and it doesn't solve it. It asks what you've tried. It asks you to walk through your reasoning. When you stumble, it asks a smaller question. It's a tutor that won't hand you the answer because handing you the answer would defeat the entire purpose. Some students hate this for the first three sessions and then notice their grades moving.

If you want to be told the answer, that's what Google is for. If you want to learn the material, you have to do the explaining.

A note on technique vs. system

Every one of these techniques has been in print since the 1980s. The reason students don't use them isn't ignorance. It's friction.

Retrieval practice requires question banks. Spaced repetition requires scheduling. Interleaving requires a coherent picture of all your courses at once. The Feynman technique requires a partner who'll push back without giving up the answer.

Each one, on its own, is a project. Together, they're a part-time job that nobody who's also taking 18 credits is going to maintain past week three.

What changes when you have a system that does the organizing — syllabus to topic map, calendar to weighted schedule, notes to quizzes and cards, daily plan that mixes them — is that the techniques become the default. You don't decide to interleave. You just open the plan. You don't decide which cards are due. You just review the ones that show up.

The techniques don't get easier. The friction around them disappears. That's the whole game.

What a typical week ends up looking like

A second-year engineering student with five courses, a Friday thermo midterm, and a Tuesday lab report:

Sunday evening, 10 minutes. Drop the lab report deadline into the calendar. Fennie reweights the week — Monday and Tuesday lean toward thermo and the report, Wednesday onward leans heavier on thermo as the midterm closes in.

Monday morning. Open the plan. Three items: a 25-minute retrieval pass on thermo entropy from last week's notes, a quiz on diff eq concepts that haven't shown up in a while (interleaving), 15 minutes on a flashcard set tagged for the lab. Total: maybe 70 minutes.

Tuesday. Lab report due. Fennie's plan is lighter on new material, heavier on a review that pulls thermo back in front.

Wednesday through Thursday. Thermo dominates the plan. The chat tutor walks through a problem set without solving it. Two flashcard reviews. A quiz built from his own notes that targets the specific subtopics his memory profile flags as shaky.

Friday morning, before the midterm. A 15-minute warmup pass on the highest-yield items.

He didn't decide any of that. He decided what to put on the calendar. The system did the rest.

Where to start

If you're going to change one thing this week, change retrieval practice. Stop rereading. After every reading session, close the source and write down what you remember. Check it. The first few times will feel humiliating. That's the technique working.

If you want the rest of the loop on autopilot, that's what Fennie is for.

Try Fennie