01
Import
A PDF, a photo of your handwritten notes, a screenshot of the lecturer's slide, or text you paste. Whatever the course actually gave you.
Dyscendi — the art of learning
Your notes become
your cards.
Snap a page of your course. Dyscendi drafts the flashcards, in order, for you to approve. Then it brings each card back the moment before you'd forget it — not sooner, not later.
Free to start · iPhone and Android · Built for students 13 and up · English and French
The problem
Everyone says make flashcards. Nobody mentions the evening it costs.
"I have 200 pages to review and I don't even know where to start."
"Flashcards work — but making 80 by hand costs me a whole evening."
"I ace the midterm and a week later it's gone for the final."
Re-reading your notes feels like studying. Mostly, it isn't.
How it works
Three steps, in this order.
02
Approve you decide
The AI proposes cards, ordered from foundations to detail. You read them, fix them, delete the weak ones. Nothing is saved until you say yes.
03
Review the scheduler
Each card returns on the day you're about to forget it. You rate how it went; the schedule adjusts. Turn on review reminders and your phone tells you when your cards are ready. One deck at a time.
An evening of card-making becomes about a minute.
The method
Built around how memory actually works.
Your brain lets go of new things on a fairly predictable curve. Dyscendi's job is to catch each card just before it slips.
The scheduler medical students rely on
Dyscendi uses FSRS-6, the current generation of spaced-repetition scheduling. It tracks three things per card — how long the memory lasts, how stubborn that card is for you, and where you stand today — then picks the day a review pays off most.
In the public FSRS benchmark, FSRS-6 predicted recall more accurately than half-life regression — the model Duolingo published — for 98.3% of the 9,999 study collections tested.[8] That's their measurement, not ours.
You set the standard
A target-retention setting in plain language, with its trade-offs spelled out. Quiet term, or finals in two weeks? Move it.
Target retention
The evidence, in short
None of this is our discovery. It's a century of memory research, and Dyscendi is built on three of its sturdiest findings.
- Forgetting follows a curve. Ebbinghaus charted it in the 1880s. A 2015 replication found nearly the same numbers: about a fifth of the learning effort was still saved a day later.[1]
- Spacing beats cramming. A review of 317 experiments found that the longer you need to remember something, the longer the gap between reviews should be.[2]
- Testing yourself beats re-reading. A week after studying, students who had tested themselves recalled 61% of a text. Those who had only re-read it recalled 40%.[4]
| Study | What was tested | Result |
|---|---|---|
| Roediger & Karpicke, 2006[4] | 180 undergraduates studied a text four times, or once followed by three recall tests. | One week later: 61% recalled after testing, 40% after re-reading. |
| Karpicke & Roediger, 2008[5] | Students learned 40 Swahili–English word pairs, with or without repeated testing. | One week later: about 80% recalled with repeated testing, 36% and 33% without. |
| Cepeda et al., 2008[3] | More than 1,350 people learned facts, reviewed them after a gap, and were tested up to a year later. | The best gap grew with the delay: about 20–40% of a one-week delay, about 5–10% of a one-year delay. |
| Dunlosky et al., 2013[6] | A review of ten study techniques students commonly use. | Only two earned the top rating: practice testing and distributed practice. Re-reading and highlighting were rated low. |
| FSRS benchmark[8] | Scheduling algorithms predicting recall across 9,999 study collections and 349,923,850 reviews. | FSRS-6 beat half-life regression in 98.3% of collections. Newer and larger models beat FSRS-6. |
Neuro-ergonomics: designed for the brain, not against it
The method only helps if the app stays out of your way. Every screen is built to keep cognitive load low, so the effort goes into remembering — not into operating software.
- One decision per screen. The interface never asks for two things at once.
- Type chosen for reading. Contrast checked to WCAG AA across 58 colour pairs.
- Motion only when it means something. Everything respects Reduced Motion.
- A broad-daylight mode. For studying outside — because your phone doesn't know you're in the sun.
Who's in charge of what
Three parts. Clear lines between them.
You
Every call that matters
Approve or reject each card. Edit wording. Set your retention target. Nothing reaches your library without you.
The AI
Reads only your course
It works from the document you imported — not the internet. It orders cards foundations-first, cites the page each one came from, and returns fewer cards rather than inventing.
The scheduler
Picks the day
FSRS-6 decides when each card comes back and adapts to how you rate it. That's the one part that runs on its own.
Straight talk
What Dyscendi doesn't do — at least not yet.
You'd find this out in a day. Better you hear it from us.
- No offline mode. It needs a connection to work.
- No sharing. Your decks are yours alone, for now — no public decks. You can export them as a CSV file, which Anki imports.
- No search across cards yet.
- You review one deck at a time. There's no combined session across decks.
- The scheduler isn't tuned to you personally. Dyscendi runs FSRS-6 with its standard settings. It doesn't yet refit them to your own review history.
- No study of our own. The research we cite was done by others, on other learners. We haven't measured Dyscendi's effect on grades.
Everything above is on the list. This page changes when the product does.
Fair questions
"Can't I already do this?"
vs. a chatbot
A general chatbot gives you the subject. It doesn't tell you when to see a card again, doesn't keep your history, and fills gaps when your source is thin.
Dyscendi answers only from your document, and schedules the rest.
vs. Anki
Anki is excellent, and it leaves you building every card by hand after an evening of setup.
Dyscendi keeps the scheduling and takes the card-making off your hands.
Two languages
Your app in your language. Your cards in your textbook's.
Run the interface in English or French. The cards follow the document — an English chapter produces English cards, whatever language the app is set to. For anyone reading course material in a second language, one small daily friction just disappears.
The offer
Try the whole chain first.
The free tier isn't a stripped-down demo of one feature — it's the entire flow, in a small monthly amount.
Free
The whole thing, every month
20 AI-drafted cards a month, from PDFs of up to 10 pages.
- Unlimited cards written by hand
- Every review, on schedule, with no cap
- Your cards stay yours, subscribed or not
Then decide.
Premium
No limit on AI cards
Unlimited AI-drafted cards, from PDFs of up to 100 pages — a whole textbook, or exam season.
One plan, billed monthly or yearly through the App Store or Google Play, which show the price in your currency. Cancel anytime in your store account.
No ads. Not now, not later.
A deliberate choice — partly because a good share of the people using this are under 18.
FAQ
Before you ask.
Can't I just use a free chatbot?
You can get cards that way. What you don't get is a schedule that tells you when to see each one again, a history that carries across the term, and a limit that stops it inventing when your source runs thin. That scheduling is most of the value.
Isn't Anki free?
It is, and it's very good. Dyscendi uses the same family of scheduling algorithm. The difference is that Anki leaves you to build every card yourself; Dyscendi drafts them from your notes and you approve.
Won't the AI feed me nonsense?
That's why it isn't allowed to write into your library. It only proposes. You read every card before it exists, and each one names the page it came from, so you can check it in seconds. If a document is too thin for the number of cards asked for, it returns fewer rather than making things up.
Is there research behind this?
Behind the method, yes: spacing and self-testing are among the best-supported findings in the study of learning. Behind Dyscendi specifically, no — we haven't run a study of our own. The science page lists each source and says what it does and doesn't show.
Will it remind me to review?
Yes. Turn on review reminders — the app offers it during a session, and the switch is in Settings › Review — and your phone notifies you when the cards you rated Again, Hard or Good are ready to be reviewed. Reminders are off until you turn them on, and you can review at any time without waiting for one.
What does "Dyscendi" mean?
It's a coined spelling of ars discendi, Latin for "the art of learning". Said /dis-SEN-dee/. The name points at what you came for, not at the tool.
Where do I get it?
Dyscendi is on the App Store for iPhone and on Google Play for Android. It's free to start; no card details are asked for.
Get the app
Start with your next chapter.
Free to start, on iPhone and Android. Questions? There's one inbox, and a person reads it.
