Free · Bring your own chatbot · No API keys
You don't learn by reading.
You learn by remembering.
APLENDE turns any subject into a real course — then makes you produce it from memory and schedules every review for the day you were about to forget.
Works with Claude, ChatGPT, or Gemini — you paste the prompt, it writes the course, the app runs the method.
Rereading feels like learning. It isn't.
The most replicated finding in learning science is also the most ignored: the study method that feels productive is the one that fails. Fluency is not knowledge.
| Method | How it feels | Retention after a week |
|---|---|---|
| Read it three times | Smooth, fluent, productive | Poor |
| Read once, recall three times | Effortful, uncomfortable, slow | ~50% better |
Roediger & Karpicke, 2006 — and every replication since.
How it works
01
Generate a class
Say what you want to learn, how deep, and what you honestly already know. Copy the master prompt into whatever chatbot you already use. Paste the JSON back — it's validated before a single row is saved.
02
Read it once
Lessons are written to be read, not highlighted. Rereading is the trap this whole app is built to keep you out of.
03
Prove it
The recall panel asks you to produce the answer from memory, and makes you commit to a confidence rating before it reveals anything. Then you grade yourself: Again, Hard, Good, Easy.
04
Come back at the right moment
Every lesson, card, and concept gets scheduled for the day you're about to forget it. Sessions mix your classes instead of letting you binge one.
What you actually do
Not a video course. Not a chat window. A study loop with one uncomfortable question at the center of it.
In your own words, what is the testing effect and why does re-reading feel better but work worse?
You commit to a confidence rating before the answer appears.
Calibration — confidence vs. actual accuracy
Flat at the top = overconfident. Now you can see it.
Mastery bars and a concept map instead of a progress percentage.
The science, and what it changed
Six findings do most of the work. Each one is a design decision here, not a blog post.
Retrieval practice
Producing an answer beats reviewing one — by a lot.
in aplende The default action after a lesson is a recall prompt, not a checkbox.
Spaced repetition
Memory decays on a curve, and each successful recall flattens it.
in aplende FSRS schedules lessons and concepts, not just flashcards.
Interleaving
Mixing topics beats studying them in blocks.
in aplende Sessions round-robin across your active classes automatically.
Desirable difficulty
If studying feels smooth, it probably isn't working.
in aplende Pre-tests you're meant to fail, optional delayed feedback, and a ~85% success target.
Calibration
Knowing what you don't know is a separate skill from knowing it.
in aplende Confidence before every reveal, plotted against your real accuracy.
Mastery, not completion
Bloom's mastery loop: you don't move on until it's actually learned.
in aplende A lesson counts as mastered after 3 good recalls on 3 separate days, plus ≥80% on its quizzes.
Ebbinghaus · Roediger & Karpicke · Bjork · Rohrer & Taylor · Wilson et al. 2019 · Bloom · Kapur
What it refuses to do
No streaks, XP, or badges
Extrinsic rewards train you to do the minimum. There's no leaderboard here — there's a calibration curve.
No learning styles
Visual-versus-auditory learners is debunked. We don't ship features built on it.
No “% complete”
Completion is a lie you tell yourself. You get % mastered, and it's harder to earn.
Don't start from scratch
Take a class someone already built. You get your own private copy — same content, zero progress, scheduled around your own reviews.
Free. No API keys. No subscription.
You already pay for a chatbot
APLENDE is the method wrapped around it. Bring whichever model you like — the strong ones write better courses.
Copy-paste in, copy-paste out
There's no integration and no key to hand over. Nothing is transmitted to us, because there's no us in the loop.
Your classes are yours
Content lives in your own database, and anything you publish is opt-in and instantly reversible.
Start with one class.
Pick something you've been meaning to actually learn. Ten minutes from now you'll be answering your first recall prompt from memory.